Slashdot Mirror

← Back to Stories (view on slashdot.org)

XenServer 6.2 Is Now Fully Open Source

Posted by timothy on from the boundless-bounty dept.

First time accepted submitter Jagungal writes "Although the core Xen hypervisor has always been open source from the start, Citrix have now released the next version of their XenServer including all features and tools under an open source license. This includes also introducing a new XenServer.org community portal. The major change for users is that they now get all features from the licensed version for free but unless they pay for support, they have to do all security updates manually. Change logs for the new version 6.2 can be found here. It's been a few years since Citrix started giving it away, free as in beer.

55 of 86 comments (clear)

  1. x86 protected mode by w4r0nc0re · · Score: 1

    Yay! now we get to port it to VCPI

  2. we ditched vmware for xenserver 2 years back... by Anonymous Coward · · Score: 5, Interesting

    and it was the best choice we ever made.

    live migration is free (as in beer). and it runs its little heart out with no problems.

    2 years now, 30 TB of files, 40 GB of mysql data, about 30 VMs on 4 hosts. not one single problem.

    The only issue we've run into is getting fully paravirtualized FreeBSD. It is a rather involved process. But once you have one VM you just copy it like a template. And luckily ZFS On Linux is starting to be good enough so we don't have to really care about FreeBSD so much.

    Plain-vanilla Xen (not Xenserver) with DRBD (et al.) making instant failover is pretty awesome too.

    Fuck VMWare.

    1. Re:we ditched vmware for xenserver 2 years back... by Anonymous Coward · · Score: 1

      You're not VMWare's target market. If you or any of your co-workers who handle Xen leave your company (or get hit by a bus/wiped out by vengeful spouses, etc.), how are they going to get support for Xen? All of a sudden your cheap IT budget explodes in one single incident, hiring someone (or paying through the nose for support) to rebuild/recover from scratch. Some businesses would basically tank at this point.

      While VMWare is pricey - I have yet to worry about any problems for the company I work for arising from the few of us knowledgeable leaving. That is called responsible IT management. VMWare support, in the few cases we have needed it, has been top notch. If I give notice tomorrow, vmware support will carry the company through any issues that would arise until they could bring someone else on board.

    2. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1, Interesting

      and it was the best choice we ever made.

      I went the opposite direction. We had been using exclusively Xenserver.... then at some point VMware released a free edition of ESXi; this is huge, and we tested it and found many advantages.

      Eventually, we ditched Xenserver. For the past few years we have been using VMware vSphere, and ditching Xenserver was one of the better choices we ever made.

      3 years now. 20 Tb of files; 6 TB of Exchange mailboxes, 500 GB of SQL Server and MySQL data, >1000 transactions per second , 16 to 1 consolidation ratio, with CPU, Memory, and Storage heavily oversubscribed; 280 VMs on 3 hosts, and no issues..

      See? Other people can do that too...

    3. Re:we ditched vmware for xenserver 2 years back... by niftydude · · Score: 4, Insightful

      3 years now. 20 Tb of files; 6 TB of Exchange mailboxes, 500 GB of SQL Server and MySQL data, >1000 transactions per second , 16 to 1 consolidation ratio, with CPU, Memory, and Storage heavily oversubscribed; 280 VMs on 3 hosts, and no issues..

      See? Other people can do that too...

      Sorry dude, this is slashdot. You lost your epeen contest with the op when you admitted your organisation uses exchange and sql server.

      --
      You can never know everything, and part of what you do know will always be wrong. Perhaps even the most important part.
    4. Re:we ditched vmware for xenserver 2 years back... by Anonymous Coward · · Score: 3, Insightful

      Just like there are many people on the planet that know VMWare, there are many people on the planet that know Xen. Just like you didn't learn VMWare by being born with the knowledge, there are manuals for Xen too.

    5. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1, Flamebait

      Sorry dude, this is slashdot. You lost your epeen contest with the op when you admitted your organisation uses exchange and sql server.

      Would it be a good time to remind you that there aren't really any open source e-mail server products that are truly enterprise grade?

      Sendmail doesn't work so well, when you are required to allow users to keep 20 Gigabytes in their mailbox, Share calendars / Tasks; access mail on their iPhone or Android device, Spamassassin's spam filtering sucks even with Bayesian filtering (due to the low kill rate and high false positive rate, and high CPU cost of Spamassassin),
      and POP3/IMAP connectivity has security issues -- regarding possibility of corporate data being exfiltrated, and inability to remotely wipe mobile devices connecting with POP/IMAP, and ActiveSync is indeed up to the challenge?

      Furthermore... i've yet to find an open source webmail platform that works with AD RMS -- Rights management services; or Windows IRM ( Digital Information Rights Management ), for encrypting sensitive corporate e-mail to ensure that it is not accidentally leaked outside the organization.

    6. Re:we ditched vmware for xenserver 2 years back... by Anonymous Coward · · Score: 2, Insightful

      Since you mention sendmail and email box sizes, you obviously don't know WTF you're talking about. Despite what you're learned working with exchange, not all MTAs have be the same giant POS as your "mailbox."

      Also, bitching because there's not an OSS platform that doesn't work with Windows proprietary solution doesn't mean that there's a problem with OSS.

    7. Re:we ditched vmware for xenserver 2 years back... by Anonymous Coward · · Score: 1

      Fewer who know Xen than VMWare, though. We've been trying for 3 years to hire a Xen expert to replace the prior one -- who is still at the company, but who got promoted to management and doesn't have time to keep Xen up -- and are increasingly leaning towards ditching XenServer for VMWare just because nobody except this guy understands it.

    8. Re:we ditched vmware for xenserver 2 years back... by niftydude · · Score: 1, Insightful

      Would it be a good time to remind you that there aren't really any open source e-mail server products that are truly enterprise grade?

      At least you didn't go so far as to try to defend MS SQL Server. Who mentioned OSS? I didn't. The reason the MS unholy trinity of server services (by which I mean exchange, sql server and IIS) are immediately disqualified for me is that they all need a GUI based operating system. Something about that just makes me want to run screaming while waving my hands in the air. If you are doing anything serious (or even moreso if you aren't), you generally don't have the RAM to waste on a bloated operating system that has to start up and maintain a GUI whether you are using it or not. You want to cut down on pointless overhead - you want a server which is a true headless server.

      But while we're on the topic, postfix can be set up with folders, not a monolithic mailbox file, and so I'm morally certain it will handle bigger inboxes than exchange does. POP3/IMAP has security issues, but so does exchange. PGP encryption is still better than anything out there. AD RMS is only relevant to people who want to use windows proprietary stuff, so not at all. And anyone trying to get sensitive information off portable devices knows that to avoid remote wipe, all they have to do is put the thing in flight mode - so none of those remote wipe solutions are worth shit IMHO.

      I will give you the fact that exchange does better calendar stuff than anyone else, but try integrating that into someone's android smartphone calendar. Either it doesn't work, or the phone manufacturer specific solution is such a battery hog that it isn't worth it.

      --
      You can never know everything, and part of what you do know will always be wrong. Perhaps even the most important part.
    9. Re:we ditched vmware for xenserver 2 years back... by msh104 · · Score: 4, Interesting

      The Zimbra Open Source Edition is probably a very good choice.
      - 99% of all companies don't need features then the open source edition.
      - it supports large mailboxes very well. ( some of our employees have 21gb mailboxes, it still runs smooth )
      - You can buy a plugin for encryption if you really need it.
      - Mail (IMAP), Calender (iCal) and adressbook (LDAP) sync is possible to almost any device.
      - You can always get the commercial version if you need the extras.

      I don't think you can remotely wipe your mail using an open source product but nowadays you might simply get any android of iphone device and use a wiping app. Maybe not as convenient but it works.

      Spamassasin can work very well ( it certainly does for us ) using external blocklists and distributed mail analysis services ( dcc, razor2 ) in addition to it's core filters. We added greylisting as well. Everything runs as part of the Amavis product. We don't use Bayesian filtering though. While good on paper we found it to be to unpredictable in real life. ( people reporting valid mailing lists as spam instead of unsubscribing, etc ) Instead we added around 15 additional custom spam filter lines over the years but that's it. Now all our spam is gone. We filter mail for over 1500 domains and our customers have never been happier.

    10. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 4, Insightful

      . If you are doing anything serious (or even moreso if you aren't), you generally don't have the RAM to waste on a bloated operating system that has to start up and maintain a GUI whether you are using it or not.

      So there is a bit of this RAM used to provide GUI functionality. It's worthwhile, because it means some maintenance tasks to junior staff whose skills for picking up CLIs are more limited; on the other hand, they can still be trained to manage the server, using remote management tools of course.

      There is practically no reason an administrator should be logging into the server and starting up a graphical console, since all administration tools can be installed on their workstation and used remotely.

      Furthermore... in 2008, core install was introduced, which no longer includes a GUI for servers, and this is supported with SQL 2012

      In Windows Server 2013; with some exceptions, the desktop experience is not required on servers, and generally, there will be no GUI.

      Anyways... the success of a hypervisor should not be judged based on the perceived quality of the applications it has virtualized. It is not a more meaningful feat to run MySQL in a hypervisor than it is to run MS SQL in a hypervisor.

      If anything.. with MySQL there are fewer sizing hints, AND the operational metrics provided by the database engine are much sparser than the detailed instrumentation that MS SQL provides -- with MS SQL, you get a heck of a lot better information about the performance and sizing.

      At least you didn't go so far as to try to defend MS SQL Server. Who mentioned OSS? I didn't.

      MS SQL server is the only backend supported by some applications, and some developers.

      Personally, I would favor Oracle, but getting anyone to agree to pay for it, is a problem.

      The fact of the matter is SQL server provides robust hitless failover clustering functionality. Postgres and MySQL do not provide this; although they are getting closer. They are worlds apart in terms of features, so it's not really fair to pick one or the other as a dilemma play, now is it?

      Some application owners will demand MS SQL, and some will demand PostgreSQL, and that's OKAY.

    11. Re:we ditched vmware for xenserver 2 years back... by Eunuchswear · · Score: 5, Informative

      It's not MTA functionality. It is LDA functionality.

      Sendmail writes a flat file... this results in some limitations

      No it doesn't

      define(`confLOCAL_MAILER', `cyrusv2')

      Problem solved.

      --
      Watch this Heartland Institute video
    12. Re:we ditched vmware for xenserver 2 years back... by gl4ss · · Score: 2

      You're not VMWare's target market. If you or any of your co-workers who handle Xen leave your company (or get hit by a bus/wiped out by vengeful spouses, etc.), how are they going to get support for Xen? All of a sudden your cheap IT budget explodes in one single incident, hiring someone (or paying through the nose for support) to rebuild/recover from scratch. Some businesses would basically tank at this point.

      While VMWare is pricey - I have yet to worry about any problems for the company I work for arising from the few of us knowledgeable leaving. That is called responsible IT management. VMWare support, in the few cases we have needed it, has been top notch. If I give notice tomorrow, vmware support will carry the company through any issues that would arise until they could bring someone else on board.

      you can't buy support for xenserver? because that's what you're saying. unless you actually work for vmware, because you know, nobody really uses the termp top notch unless they're writing from a script or just trolling. vmware is just bullshit fleecing licensing.

      --
      world was created 5 seconds before this post as it is.
    13. Re:we ditched vmware for xenserver 2 years back... by drinkypoo · · Score: 2

      If you are doing anything serious (or even moreso if you aren't), you generally don't have the RAM to waste on a bloated operating system

      Nonsense. How much RAM does the most bloated operating system use? Less than a gigabyte? How much does RAM cost now? How much RAM can you get into a PC server now? Oh, so that hypothetical gigabyte is basically irrelevant? I see.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
    14. Re:we ditched vmware for xenserver 2 years back... by gbr · · Score: 1

      I've been using XenServer for five years (3 servers, 23 VM's, 5TB on iSCSI).

      I'm seriously considerring dumping it for KVM (Proxmox, specifically).

    15. Re:we ditched vmware for xenserver 2 years back... by Neo-Rio-101 · · Score: 1

      Well for starters, ditch Sendmail, use Exim, and then implement DRBD with linux HA, Dovecot, and Perdition.
      If by enterprise grade, you mean "good enough for an ISP to use", then open source does just fine with a bit of tweaking and fine tuning.

      So spamassassin uses a lot of CPU? Have you seen the resources an exchange DAG requires?
      Granted, exchange has a nice calendar system and the extra goodies that businesses want, but that's a real cost/benefit analysis given licensing and the amount of hardware you have to throw at it.

      --
      READY.
      PRINT ""+-0
    16. Re:we ditched vmware for xenserver 2 years back... by riondluz · · Score: 2

      I managed a zimbra system some time back and it was OK, support was pretty good. But it was all for the outlook plugins. Too much overhead otherwise (imo). Citadel is another good product.

      Still, when all is said and done, count me in as a fanboy of sendmail on a xenU; despite my peers always singing the praises of postfix.

      I use it with selinux, the milters, razor, pyzor, dcc, clam, combined with a virtusertable that routes non-existant users to the pit:
      @some-domain.com error:nouser No such user here
      and requiring reverse-dns for connecting hosts
      (tip of hat to
      http://h30499.www3.hp.com/t5/Messaging/sendmail-refuse-mail-with-no-reverse-lookup-PTR-record/td-p/3194706 )
      I've had no issues and nearly no spam, in years of operation.

      My only xen beef is my stupidly creating a xenU on an LVM volume that's too small w/too little storage. This resulted in having to NFS-mount (or DRBD) a yum-cache as well as a clam download dir.

      --
      resist propaganda
    17. Re:we ditched vmware for xenserver 2 years back... by amginenigma · · Score: 2

      I run in what I'd say is a small shop, we have 21 sites the smallest of which have two hosts 13 'guests' with 12TB of shared storage. Our average site has four hosts, 22 'guests' and 48TB of storage (almost ready to double those as they are nearly full). The two data centers each have 11 hosts, over 100 hosts (virtual sprawl I'll admit I'm losing count and that's a bad thing) and 120TB of storage each. For VMWare with that many nodes we were looking at hundreds of thousands of dollars that no one wanted to spend. We switched to XEN roughly two years ago, in that time I've had one issue that I was literally able to 'google' the answer to before our Linux vendor responded to the support request. Yes you can get support for XEN from any of the big guys, and guess what, at least for the distro we use support for XEN is included. For us VMWare couldn't touch that with a ten meter cattle prod. Oh I manage all of that with one 'console', and no not one per site, one console to rule them all (ha had to sneak that in there). Was it free as in beer, sure, but that caused us to learn what we run rather than just poke buttons and call for support the instant something (anything) goes even slightly awry as I've seen others running competing solutions do.

    18. Re:we ditched vmware for xenserver 2 years back... by niftydude · · Score: 1

      since we're talking about VMs, when you have several instances running on your hardware, then yes, all those stray gigabytes do add up. And ms has basically admitted this is a problem, since I'm told you can run Windows server 2013 in a completely headless mode.

      So they have finally seen the need to catch up to a feature that's existed in the last 50 years of *nix and mainframe, even if you don't.

      --
      You can never know everything, and part of what you do know will always be wrong. Perhaps even the most important part.
    19. Re:we ditched vmware for xenserver 2 years back... by vux984 · · Score: 1

      If by enterprise grade, you mean "good enough for an ISP to use", then open source does just fine with a bit of tweaking and fine tuning.

      I think by enterprise grade he means that it has the FEATURES the modern enterprise user expects.

      What is typical ISP mail? POP+IMAP with a couple hundred megabyte limits. No calendar/contact sync to my phone or desktop client. No remote wipe of my phone. No Calender sharing, no global address book, no encryption, no certificate authentication,...

      The email i get from an ISP is anything but enterprise grade.

    20. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1, Interesting

      So spamassassin uses a lot of CPU? Have you seen the resources an exchange DAG requires?

      Yes. I have, and Spamassassin/Amavis is orders of magnitude more expensive CPU-wise. This is probably due to the very massive amount of spam being attempted, and the inability of RBLs and tools such as DCC and Vipul's razor to effectively divert them, before Spamassassin starts chugging away at massive concurrency.

      All the more bothersome, when some domains want "wildcard forwarding rules", so brute-force spam attempts at a domain fire off hundreds of messages.

      If by enterprise grade, you mean "good enough for an ISP to use", then open source does just fine with a bit of tweaking and fine tuning.

      No. I mean good enough to satisfy Enterprise users, and managers of Enterprises, such that they are satisfied of the features, performance, capacity, availability guarantees, and user experience that they have no benefit to running their own mail server.

      This by the way includes rapid failover capabilities of all mail-related operation, load balancing, spam filtering, and prevention of "misbehaving users or devices" from adversely effecting server operations -- (per user throttling).

      There is Non-MS, Non-GUI technology that might provide many of the things. But at the end of the day What the users want matters more. If the end user is insistent on features that can only be provided by a GUI-based server solution, then that is probably what they will be getting.

      Well for starters, ditch Sendmail, use Exim, and then implement DRBD with linux HA

      Not happening. Well-justified hate relationship developed with Exim over the years, due to MTA induced failure after failure on large shared hosting deployments, poor logging, and general lack of simple configurability -- I think perhaps the only Linux mail server software to suck more than Exim is QMail; they are pretty much at the same level.

      Postfix, on the other hand is great.

      DRBD with linux HA

      The problem with DRBD with Linux HA; other than the fact that only a unix guru can maintain it and safely execute a disaster recovery (which 98% of the engineering staff are not qualified to do, and the portion that think they are qualified but are not are the most dangerous), is that DRBD doesn't meet acceptable standards of robustness, and requires manual attention in the event of a failover, and for failback: it's active/passive, and failover can really only be leveraged safely in the event of a catastrophe.

      In other words: DRBD in its current form is only really suitable to be used as disaster recovery type system, not a host failure response type system.

      In testing DRBD has proven to be unreliable in various failure modes, we would wind up with damage to data caused by DRBD itself, or operator error trying to get the failback to work like it's supposed to after a split-brain incident. Also, anything reliant on fencing by "killing" another server externally is not acceptably sane -- robust clustering systems utilize self-fencing nodes and provide additional methods of making quorum.

    21. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1

      since we're talking about VMs, when you have several instances running on your hardware, then yes, all those stray gigabytes do add up.

      The Windows GUI is megabytes, not gigabytes, and these memory pages are transparently shared between VMs running on the same host, so the memory for GUI code in memory is deduplicated and "used once" for the benefit lots of VMs it's not that bad.

      Most of the memory waste is due to the .NET framework, which leverages Garbage-collection based memory management for both server side and client side applications.

      Java has a very similar issue with its Garbage-collected heap.

      There is a GB or so of memory per server needed for filesystem cache, a small amount of memory required for server applications, for SQL servers SQL Engine cache memory, and then, for all kinds of clients and servers: many gigabytes for .NET or Java heaps.

      Futhermore, the .NET or Java garbage collected heaps do not overcommit as well as "OS GUI memory" or resident Application code memory, because the Cache memory represents data that is specific to a workload, there are fewer memory pages in common.

      The bulk of the memory usage is the result of inefficient application coding, or the use of inefficient Software development frameworks such as .NET, Java, Python, Ruby, Perl, PHP, etc... instead of C++ or Objective C, Not an inefficient operating system.

    22. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1

      Think of it this way: how much unused memory is going to waste in a server that only hosts one instance of something?

      The answer is a whole lot, because application owners and novice sysadmins very frequently overestimate or intentionally overstate memory and CPU requirements, thinking more is always better.

      So rightsizing, requiring justification for resource allocation, and time spent tuning and studying possible $/Compute improvements are important parts of capacity management.

      But page sharing, overcommitment, ballooning, and swapping, as implemented are an extreme help.

      "Dynamic memory" technologies such as that in Xen / Hyper-V to a much more limited degree.

    23. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1

      Also, bitching because there's not an OSS platform that doesn't work with Windows proprietary solution doesn't mean that there's a problem with OSS.

      No. I am just disappointed that there is no OSS platform that is suitable, because I really want there to be. I would love for all the code required to implement Enterprise services to be open source, and for them to be scalable, proven, robust, self-maintaining, and not requiring ad-hoc script work and daily attention from highly experienced unix engineers.

      I'm not bitching that there's not a satisfactory OSS platform for Enterprise mail, and that the OSS hypervisor's available are not the most cost-effective choices: I am just explaining that is the world we live in, that we are to accept, that causes the decision, and the inherent characteristics of what is available right now, and what Enterprise users demand That make no OSS solution the right choice in some cases.

      Because the OSS options are missing the last 20% of the characteristics Enterprises require, even though the other 80% is there.

    24. Re:we ditched vmware for xenserver 2 years back... by deek · · Score: 1

      I think what you mean to say is the default MDA (or LDA) agent for sendmail writes a flat file. It looks a bit strange when you claim sendmail also has MDA duties.

      Unsure why you think a flat file cannot exceed 2GB. Of course, it's not optimal to have a mailbox file that big, but if you're running a mail system that deals with large mailboxes, you'd have switched to the Maildir format years ago. This also helps with mailboxes with a very large number of messages, or at least, it pushes the problem to the filesystem. You can then pick and choose a filesystem, and tune it for a large number of small files.

      Also unsure why you list Dovecot and Cyrus as a mail/IMAP client. I think perhaps you mean IMAP server, but you also mention Pine, which is a client.

      With a bit of effort, you can scale these services over as many servers as you wish. Have a look at http://www.linuxjournal.com/article/9804 for an example. It's an old document, but is still relevant for this type of design.

      You're right that there's no push email feature for open source services. The closest I know of is using the IMAP IDLE feature, which keeps a connection open to the mail server. The mail server then notifies the client immediately through this connection. Not quite push email, but comes very close.

      Unsure what you mean by enforcing security policy.

    25. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 2

      Unsure what you mean by enforcing security policy.

      I'll give you some examples, with how this is achieved within an Exchange environment in a typical enterprise:

      • Users outside the enterprise access their mail through Activesync or OWA (Outlook Web Access)
      • These services are published to the internet by a Forefront UAG or TMG -- smart card, or password and 2-factor access token are used for login via OWA, so this is secure.
      • Activesync is the open standards-based protocol utilized with smart phones, in order to synchronize things from a mail server to the mobile device.
      • When an enterprise authorized smart phone is provisioned, a client-side certificate is installed on the device, to allow Activesync connection using SSL, so again, there is two factor auth
      • Activesync devices, when they associate download a security policy; this provides mechanisms that can be used to enforce policy
      • Example: PIN Required policy, requires that the user must set a PIN on their smartphone, and enter their PIN number to gain access to Enterprise mail. In addition, after repeated failed entries, the device will be wiped.
      • Aside from activesync policies; the mail server the device is associated with gets a Remote Wipe Function; in the event that the smart phone is lost or stolen, the enterprise can push down a message that will cause all the sensitive e-mail to be wiped; the phone will reset to factory defaults.
      • OWA, Outlook, and Activesync protocol devices support meeting requests; free/busy availability for other people; calendar sharing and appointment functions. There are many tools provided through open standard activesync that are not available on generic IMAP clients.
      • OWA and Outlook support functionality to Classify Sensitive Messages, view encrypted message, and send encrypted messages and documents, which can be disseminated only within the enterprise -- in other words, they cannot be accidentally forwarded outside the organization. So called Digital Rights Management or Information Rights Management functions. The closest OSS equivalent is PGP and does not provide a user friendly interface for end users, that can be readily administered by IT security.
      • Aside from rights management, there is this concept of Data Leak Protection; where messages that contain sensitive information, will be detected by automated algorithms on the mail server, and users prevented from forwarding the messages outside the organization ---- this provides robust security against accidental leaks since Rights management services protected encrypted documents cannot be viewed, except by valid users on computers with a user certificate and computer certificate that acquired a "view license", using mail software that supports RMS --- generic POP/IMAP clients, and generic document viewers do not support implementing this kind of security.

      With a bit of effort, you can scale these services over as many servers as you wish.

      This is only horizontal scaling, and is inefficient, meaning that more overall resources and cost is required to scale up massively -- DESPITE the fact that the cost of each license is $0 with OSS; additional computers and electricity happen to be very expensive. The available OSS does not have adequate vertical scaling, and would require lots of custom bespoke system customization.

      Unsure why you think a flat file cannot exceed 2GB. Of course, it's not optimal to have a mailbox file that big,

      Flatfiles are not inherently restricted, but on many systems there is no largefile support. The exact details are murky, but there are plenty of reasons a 2GB mailbox doesn't work out so well. I am really using the example to highlight an entire class of issues.

      You'd have switched to the Maildir format years ago

      You m

    26. Re:we ditched vmware for xenserver 2 years back... by X0563511 · · Score: 2

      This is just a guess, but you're probably not offering enough pay.

      --
      For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
    27. Re:we ditched vmware for xenserver 2 years back... by deek · · Score: 1

      Thanks for the clarification. Some good information there, but still not as coherent as I'd like.

      Excellent points about the security policy. Other than Activesync also possible to use with open source software, I can't think of any equivalent ability for the other features you mention.

      I'm confused as to why you think OSS doesn't have adequate vertical scaling. Add more RAM, you can run more simultaneous connections. Add more/faster disks, you can fetch mail more quickly. Add more CPU, and, well, it's just as responsive. Mail servers aren't cpu bound, unless you have a fairly underpowered cpu.

      Gosh, the old 2GB file size limit on 32 bit systems. Haven't seen that for a decade. Probably not a good example to use.

      I quite like Maildir. I don't deal with hundreds of billions of emails, though. Just manage a mail server that stores email for 400-500 domains, with around 50,000 mailboxes, and 4 or 5 million emails. A one server job, really. Maildir works very nicely. It doesn't have an indexing system per se, but service software can implement that. Dovecot and Cyrus do just that. Works just as fast as the Exchange system we have here, which handles email for around 90 employees, although the Exchange system does suffer from the occasional Outlook connection timeout when it gets heavy with disk IO. I blame the sales people and their email searches.

      Backups from Maildir aren't too bad. I get up to 80GB/hour. After hours, of course. Via Rsync, to a remote backup server. I do have directory indexing turned on for the filesystem. Necessary when dealing with so many files.

      Maildir restores are far from impossible. Just copy the relevant restored mail files to the correct directory. Oh, and if running Cyrus, reconstruct the mailbox. Easy and convenient. Beats the hell out of having to splice the restored email into an mbox file. Haven't done that for years.

    28. Re:we ditched vmware for xenserver 2 years back... by dan_barrett · · Score: 1

      Interesting.
      We moved from Open-xchange (maildir, cyrus imap) to exchange 2010 a while back mostly for political reasons.
      Our email backups have become a nightmare.
      We can't backup exchange mailboxes while they're being replicated to our DR site - the exchange server blue-screens BY DESIGN to provent the mailstore from being corrupted. Thanks, Microsoft! After shutting dowen replication we have to backup the entire 200GB database as one blob, every time - this takes at least a couple of hours over the network to our backup server.

      The mailbox DAG loses sync fairly regularly, which means you have to dump a replication copy and start again (ie copy the whole thing over the network again.) This kills mailbox performance dramatically.

      In comparison we used rsync to replicate our maildir directories to our DR site previously (along with openldap slapd replication) - it worked great. Rsync backup took minutes to complete as it only backs up the changes.individual email deach and recovery from backup is easy, too, just put the emailback int he user's mailbox and re-index the mailbox, compared to recovering the entire mailbox database.

      Finally - Postfix appears to crap all over Exchange's MTA in terms of performance and logging/problem resolution. We use postfix in front of our exchange servers and it's regularly waiting for exchange to catch up.

      Sure, exchange has lots of end user features but i'm not impressed with the backend storage solution at all.
      Give me postfix/IMAP/maildir over Exchange any day.

    29. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1

      We can't backup exchange mailboxes while they're being replicated to our DR site - the exchange server blue-screens BY DESIGN to provent the mailstore from being corrupted.

      It sounds like a possible storage sizing issue, or an issue with the way the replication has been put into place; it's not acceptable to block an Enterprise application's production I/O for DR operations: I wonder if your replication solution is freezing writes or using up all the disk IOPS... it is definitely possible to design a deployment with Exchange databases replicated at the storage layer, without any issues.

      In fact... I have experience doing that -- I have not gotten a single blue screen on production VMware virtualized Exchange, although I have in the lab seen restarts, if there is a storage failure event lasting longer than the SCSI timeout - any server could be expected to treat such circumstances as a hardware failure.

      After shutting dowen replication we have to backup the entire 200GB database as one blob, every time - this takes at least a couple of hours over the network to our backup server.

      Use a VSS-aware backup solution with support for incremental backups.

      Microsoft has provided a backup API that provides incremental backups to the backup solution.

      You get a transactionally consistent backup that you can be assured matches the source database as of a point in time.

      Rsync does not provide the same level of transactional consistency -- in other words, things can be missing from the backup - or worse, on a busy mail server, rsync never finishes.

    30. Re:we ditched vmware for xenserver 2 years back... by pnutjam · · Score: 1

      Good point. I'm sick of seeing the constant ads posted on craigslist for top tier talent at $15 an hour. If you are a tech company, paying secretary wages is a mistake. Talent costs money.

      --
      Cheap storage VM.
    31. Re:we ditched vmware for xenserver 2 years back... by pnutjam · · Score: 1

      They guys that know what they are doing are expensive no matter what platform they are supporting. Sure there are plenty of know-nothing vmware guys since it is a more common buzzword. Too bad you are just getting a warm body that can take directions and call a support line. I good tech/admin is totally different.

      --
      Cheap storage VM.
    32. Re:we ditched vmware for xenserver 2 years back... by ae1294 · · Score: 1

      Why do you believe that all homosexuals are pedophiles and in the closet? Is it because you're a pedophile and in the closet? That would make sense -- you can only imagine the world as you experience it. Maybe your were molested as a child. Well, that's wrong and nobody should have to experience it. But you can break the chain, come out of the closet, and have normal homosexual relations with other adult men.

      But I like the closet, It makes me feel warm and safe and I was trolling the troll Mr Anonymous Coward..

    33. Re:we ditched vmware for xenserver 2 years back... by mysidia · · Score: 1

      The Zimbra Open Source Edition is probably a very good choice.

      There's a problem with that.... you do any business with the vendor of Zimbra at all, and the EULA that you have to accept forbids using the open source edition of Zimbra, even for separate unrelated business.....

      Otherwise, yeah, the open source Zimbra is a potential alternative, for people to look like who are not forbidden by software licensing terms to consider that alternative.

  3. Interesting.. by CimmerianX · · Score: 1

    I've tried Xen. It is definitely not on the same level as VMWare. Perhaps this latest free version will allow clustering similar to KVM. The only benefit to Xen is that it will load on almost any hardware.

    1. Re:Interesting.. by Anonymous Coward · · Score: 1

      Xen and XenServer are really different. XenServer uses Xen, but it supports clustering, failover, ha etc...it's really not that different from esxi in my experience.

  4. Define open source by Anonymous Coward · · Score: 5, Interesting

    According to

    http://www.xenserver.org/about-xenserver-open-source/gplv2-license.html

    the licenses used include:

    AFL
    Artistic
    ASL 2.0
    BSD
    BSD-like
    LGPL (v2+, v2.1 , v2.1+, v3+, v2+ with linking exception, with linking exception)
    GPL (v2, v2+, v3, v3+, unspecified version, v2 with linking exception)
    OSL
    MIT (v1.1, unspecified version)
    OpenLDAP
    Zlib
    PSF

    That list also includes:

    Qlogic (link is to http://www.qlogic.com/supportx0/agreement.asp , but that's borked)
    Public Domain
    pubkey (artefact; refers to GPG keys for some reason)
    Proprietary
    Distributable
    Freeware

    I'm just a simple hyperchicken lawyer from Andromeda, but in my galaxy, proprietary licenses aren't 'open source' let alone Free software licenses. Same goes for freeware, public domain, etc.

    For the curious, the proprietary-licensed stuff includes software from Brocade Communications, Citrix Systems (!), Emulex, and QLogic.

    1. Re:Define open source by storkus · · Score: 3, Informative

      Mod parent up: I searched for almost 15 minutes trying to find the exact "free software" license it was changed to, and failed. But, boy, finding how to use XenControl (which runs on winblows only, BTW) to "license" your server (apparently that's what Citrix calls a support contract now) is very easy; oh, and this "license" is per socket now rather than per machine.

    2. Re:Define open source by Gumbercules!! · · Score: 1

      You can actually run Open XenCenter, which I have successfully gotten running on a Mac (for the record, it took me hours to get this working) and it's easier than that on Linux. You can also fully control it from the shell.

      The problem with XenServer it it's *amazingly* unreliable. Like terrifyingly so. We had 11 hard drive corruptions in 6 months under XenServer 6+, all on rock solid hardware (XenServer 5 was really good, though).

      It's expensive and painful. I won't be touching it again.

    3. Re:Define open source by Anonymous Coward · · Score: 1

      Yes, you may be a simple hyperchicken lawyer, but you have a good eye. The problem is that you don't know what you are looking at.

      All the proprietary stuff is licensed to Citrix, Brocade, or QLogic, and it is most likely firmware. Proprietary software can still be licensed for distribution, but firmware is very often closed source. It is unfortunate for those who want to get very low to the hardware, but in reality, very few people demand the source code for their BIOS, firmware, and other devices.

      If you want to be a purist, I guess you can run the software without interfacing with these SAN storage systems. That's assuming that you had one of these storage systems in the first place. Still, it is hard to get very bent out of shape over these companies releasing drivers and allowing them to be distributed, just because they didn't release the source code.

    4. Re:Define open source by Anonymous Coward · · Score: 1

      Indeed, many of the proprietary-licensed items are obviously firmware, but others are ambiguous: elxocmcore, conversion-plugin, brocade-bfautil, iscli, sm-closed, xenserver-transfer-vm, and xha, to be precise.

      Whether or not all the proprietary stuff is firmware, and whether anyone other than a 'purist' should care, Citrix still shouldn't say that XenServer is "Fully Open Source" as they do in their press release because it isn't. As an aside, I wonder if they could get in trouble with the SEC for misleading statements like that.

    5. Re:Define open source by Gumbercules!! · · Score: 1

      Yeah I thought so at first - but the same hardware is fine on XenServer 5.6 or Hyper-V 2012 or... well.. anything else. It's possible there's other factors at play but I am happy to be rid of it. It was so unreliable and failures were always so catastrophic and rapid.

  5. Advanced features not free by mysidia · · Score: 3, Informative

    It's been a few years since Citrix started giving it away, free as in beer.

    They gave away what used to be called XenServer Enterprise functionality for free.

    What they don't include for free is:

    • "Hot" snapshot (Snapshot a virtual machine including its online RAM/Memory status) -- using VMware's hypervisor it's a free feature, Citrix makes you pay for it.
    • Cluster-wide robust per-target resource Storage and Network 'QoS' functionality
    • High availability. Not included with XenServer free as in beer; Citrix provides it as an addon.
    • Live migration of virtual machines between backend datastores (E.g. migration between SANs) without shutting the virtual machine down.
    • DRS-Like Workload balancing
    • True memory overcommitment -- you get a more limited technology, no transparent page sharing, no swapping via SSD RAM cache or page compression.
    • Role-based access control and AD integration for login to Xen servers
    • Resource pools with servers having different CPU versions. (Enhanced 'VMotion' Compatibility)
    • No distributed power management
    • Alarms and e-mail notifications.
    • Storage array offloaded cloning/copy/zero
    • No SR-IOV/GPU or other passthrough device support
    1. Re:Advanced features not free by Anonymous Coward · · Score: 4, Informative

      I beg to differ, your list is not correct and seems to be based upon the previous 6.1 distribution model : what is now open-sourced as XenServer 6.2 is what was build as XCP (Xen Cloud Platform)

      I am not telling that everything in your list is now free, but for example HA is there, as are heterogenous pools (I used them on XCP 1.6) and live migration.

      http://www.xenserver.org/overview-xenserver-open-source-virtualization/open-source-virtualization-features.html

  6. Excellent by Gumbercules!! · · Score: 1

    I have to thank Citrix XenServer. My company, which runs a virtual datacentre environment in Australia, used to be 100% XenServer based until version 6.1 came out. It was -so bad- and -so incredibly unreliable- and caused so many problems that I started looking around at alternatives. So technically, it's thanks to the terribleness that was XenServer 6.1 that we now run Hyper-V 2012 on all our servers and I am much happier. I used to like XenServer up until v6 but I am much happier on Microsoft's offering, surprisingly.

    (No: I"m not paid by Microsoft).

  7. Re:Still on 5.5 by mysidia · · Score: 2

    If it's dying; I am very heavily inclined to believe it's been mismanaged, sized improperly, misconfigured, or deployed incorrectly.

    UCS is rock solid. If you have issues with it, get some consultants out to audit and verify the configuration and design of the infrastructure piece by piece for adherence to best practices; and attempt to fault isolate to where the issue is.

  8. Re:Sucktrix by rubycodez · · Score: 2

    are you confused, maybe thinking about xen desktop? that's useful for normal business apps....if you're trying to do heavy duty rendering with the "thin client" model that's your problem

  9. Re:Sucktrix by symbolset · · Score: 5, Informative

    XenServer is virtualization and cloud tech, not thin client tech. You're thinking of XenDesktop and XenApp - both of which are sweet stuff if you need that sort of thing.

    XenServer being fully open source is cool because it creates a competitive environment for KVM, the native Linux virtualization solution. This competition will drive rapid adoption of technologies like PCI passthrough and partitioning of GPUs and coprocessors like Xeon Phi as well as other devices that seem to converge on what you seem to have meant to say. It will also promote technologies that pass user input back to the VM like voice, video and touch inputs, and support software defined networks. Everybody who possibly can will now integrate their devices with this. This will of course spur Microsoft's Hyper-V team to redouble their efforts. VMWare will laugh and laugh until the joke's on them, but in the mean time they'll earn great profits.

    /disclaimer: I work for a joint that plays with all these, but my opinion is my own. No stock in anybody but mutual funds. No benefit for me on any of these.

    --
    Help stamp out iliturcy.
  10. Re:Still on 5.5 by symbolset · · Score: 2

    The first iteration of UCS gear neglected the bandwidth. The error has been corrected, but early adopters still feel the pain.

    --
    Help stamp out iliturcy.
  11. ganeti by halfnerd · · Score: 3, Interesting

    Check out ganeti as well: https://code.google.com/p/ganeti/

    Features
    Ganeti provides the following features for managed instances:

    Support for Xen virtualization:
    Support for PVM and HVM instances
    Live migration support
    Virtual console (on PVM) or VNC (on HVM) to control instances
    Support for virtio or emulated devices

    Support for KVM virtualization: (from Ganeti 2.0)
    Live migration support
    Support for fully virtualized instances
    Support for semi-virtualized instances (kernel residing on the host)
    Support for VNC or serial access
    Support for virtio or emulated devices

    Recommended cluster size 1-40 physical nodes

    Disk management:
    Plain LVM volumes
    Files (from Ganeti 2.0)
    across-the-network raid1 (using DRBD) for quick recovery in case of physical system failure

    Instance disk partitioning supported from Ganeti 2.0

    Export/import mechanism for backup purposes or migration between clusters, or

    Automated instance migration across clusters (since Ganeti 2.2)

    1. Re:ganeti by evilviper · · Score: 1

      The most interesting thing about ganeti, IMHO, is the basically unlimited read speeds on inexpensive hardware, thanks to the way it operates with DRBD.

      With a SAN based solution (like almost all others virtualization solutions use) your disk I/O is limited to network speed, while even a single 7200RPM 1TB+ SATA drive can saturate a gigabit network. A SAN with just a few hard drives should saturate a fairly expensive 10GbE port. Bonding several is possible, but gets expensive very fast.

      Assuming a low-end server configuration with 4x cheap 7200RPM SATA HDDs, with just 4 VM servers connected to it, you'd only need 4 such servers to surpass the I/O of a 10GbE SAN. Double that if you've got dual bonded 10GbE NICs in your SAN controller.

      --
      Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
  12. Re: we ditched vmware for xenserver 2 years back.. by DigiShaman · · Score: 1

    Fuck you! You fascist Nazi fuck! This is Slashdot. Learn to take the good with the bad. Being able to post AC is necessary regardless who abuses it.

    --
    Life is not for the lazy.
  13. The wrong approach... by Junta · · Score: 1

    It will also promote technologies that pass user input back to the VM like voice, video and touch inputs,

    I can do all of that without dealing with the hypervisor as a middle man of any sort. Rather than using SPICE to a QXL video, I'd suggest the right answer would be a mechanism that works over IP to the hosted instance directly (e.g Xpra or RDP).

    I also find PCI passthrough less interesting (it's too coarse grained) and technologies to more intelligently access GPU/Accelerator technology in a sane way more interesting. VirtualGL is a good example but doesn't include the facet of implementing that in a way accessible from a VM or OpenCL type activity, but you get the idea.

    Phi is an interesting beast since it's really a host in and of itself and the best strategy to get what performance you can is to actually run things directly on it (e.g. ssh in). Even in that configuration, it currenlty doesn't match the current Tesla adapter performace for most workloads.

    --
    XML is like violence. If it doesn't solve the problem, use more.
  14. Re: we ditched vmware for xenserver 2 years back.. by ae1294 · · Score: 1

    Fuck you! You fascist Nazi fuck! This is Slashdot. Learn to take the good with the bad. Being able to post AC is necessary regardless who abuses it.

    Wait I have no problem free speach.. but screaming RANCID ASSHOLE, COCK, ETC, ETC over and and over is just stupid and what the GP was posted had nothing to do with anything and I think it would be funny if everyone knew who he is since apparently he is so eager to post the same cut/paste message in every article. I'm not asking for him to be prosecuted or anything like that. I just want to know the time and IP address so I can track his identity down and post his picture and all these messages on facebook or something like that...