Slashdot Mirror
With OES 2.0, Novell Moves NetWare To Linux
apokryphos writes "Novell's long journey from NetWare to Linux is finally complete, with Open Enterprise Server 2.0. Linux-Watch takes a look at the newly-released OES 2.0: 'Now, with OES 2.0, the NetWare operating system kernel, NetWare 6.5 SP7, is still there if you run it, but it runs on top of the Xen hypervisor. You can also run the NetWare services, or a para-virtualized instance of NetWare, on top of Xen with the SLES (SUSE Linux Enterprise Server) 10 SP 1 kernel. So, if you're wedded to NetWare and its way of doing things, you don't have to wave good-bye to it.'"
I could have sworn they sold this product ten years ago...
Presumably the reason you are sticking with a platform that has not really changed much in a decade is because you are too risk-averse to jump to something else. That said, is swapping out your NetWare servers with "Nu-NetWare" running on top of a Linux kernel really less risky than just switching to Linux -- or to Windows with Active Directory, for that matter? If it's taken you this long to even consider replacing those servers, couldn't you have spent some of that time constructively -- by coming up with a longterm migration strategy that would enable you to minimize risk? Seriously, I have heard some arguments why NetWare is so much "better" or "more elegant" (or whatever) than a Windows network, but these arguments usually seem to hinge on some specific minor capability. It seems to me that you can get pretty much everything NetWare gives you on a Windows network with some third-party management products, with the upshot that your platform is not obsolete.
Breakfast served all day!
Netware confirms it: Netcraft is dying.
er... wait a second
eDirectory
... but not to Ubuntu. So you can have all the Novell apps on your Microsoft network.
GroupWise
ZENworks
On the other hand, Novell has ported all of them to Windows
Anyone care to comment on how nice it is to depend upon the good will of your biggest competitor for the stability of your apps?
That said, is swapping out your NetWare servers with "Nu-NetWare" running on top of a Linux kernel really less risky than just switching to Linux
Yes it is. I can test and deploy this easier than starting fresh with anything else.
couldn't you have spent some of that time constructively
I did spend that time more constructively. The boss said "I've got other things for you to do that will actually make me money. Don't worry about something that basically works."
hese arguments usually seem to hinge on some specific minor capability
It works in Netware and I can't do it as easily on any other platform. Don't denigrate something you know nothing about.
One of the fundamental premises behind your opinion is the "constant upgrade cycle" mentality.
Is IT's job making work for itself by breaking things that work or making users/systems more productive? My boss and I both choose the latter. That's why I'm happy and work lots of very regular hours.
http://www.maxineudall.com/2010/02/should-economists-be-sued-for-malpractice.html
The reason this makes sense is mostly because of driver support. Anyone seen any major support for new iSCSI SAN devices for NetWare lately?
By moving NetWare into Xen they gain the driver support SUSE Linux Enterprise Server will have, and at the same time create an environment that makes it easy to upgrade.
To the top poster - it's not exactly easy to migrate away from a platform like eDirectory once you've committed to it, and yes Virginia, eDirectory does scale better than Active Directory any day.
Here's to the crazy ones
Since then, the iFolder project has struggled, with people leaving, some wanting to rewrite the whole thing in C again (mono had some scalability issues), etc. Finally when they've managed to put in some of the features people have been wanting (multiple ifolder-servers, encryption etc), Novell in all its wisdom has decided again to make iFolder exclusive to OES2.
That's right: if you want to setup an iFolder server with the new 3.6 features, you need to buy OES2 at the premium price Novell is asking (and besides OES2 is full of other stuff many people don't want). So for Red Hat and any other distro, 3.4 is the latest version..
That's not entirely correct. You can download the iFolder 3.6 server from: http://download.opensuse.org/repositories/home:/dl9pf/. That is linked to from the iFolder site (www.ifolder.com). Now admittedly, I don't think there has been a whole lot of community involvement with iFolder 3, Novell has mostly been doing their own thing with it. But you can get source RPM's for iFolder 3.6 from the link above, along with RPM's for Fedora and OpenSuSE/SLES. They do need to do a better job of giving information to the community about what's going on with iFolder though. The website doesn't really have much on the new version - it seems like there are just occasional announcements out of the blue. There are also several different versions of iFolder mentioned on the site (3.2 - which came with OES 1, 3.4, 3.5, and 3.6). Hopefully Novell will do a little bit better job managing iFolder now that OES 2 was released. I love iFolder and use it a lot. It's great for people with laptops that need to have access to their files while off the network, but still want to have all of their files stored on a server to share with other people or run centralized backups, etc.
I briefly played around with OES 2 in VMware last night, and it doesn't seem too bad. I haven't been able to try out too many of the new features, but the installation was pretty smooth (especially in comparison to OES v1).
The good thing about what Novell is doing is that they are making Linux a viable option for a lot of midsized companies. RH or other distros work fine for small companies or large companies that have the technical people to make the glue to put everything together. But midsized companies need something more than RHEL/Fedora would provide out of the box, and may not have the expertise to put together a home built/3rd party solutions for directory services/groupware/web based file access/etc. You can certainly do those things with RHEL/Fedora, but not out of the box. With Novell (and MS, and maybe IBM & Sun too) you can get software to do all of that that works together without having to spend a lot of time putting together bits and pieces of software from different places.
Every time you post an article on Slashdot, I kill a server. Think of the servers!
There is a good reason to keep netware on a network, if its already there, and its not just laziness or fear.
Actually, two things come to mind immediately - TCO, and the netware permissions.
From my experiences when I ran netware servers, a system could be thrown together for about $5-600 (thats hundreds,
not thousands...) that could serve directory services, files, and print jobs to 200 clients simultaneously without batting an eye, and do it nonstop for months. Its hard to get anything else to match those numbers for that little $$.
Though one of the true hallmarks of netware is the permissions set that it has, that I really haven't seen an equal to in anything else. IIRC, there were 8 different permissions that could be set in netware, as opposed to the 3 in *nix. It is particularly valuable if you want to use directory structure as part of your workflow - for example a user could have a directory where they could write, read, but not modify or delete. I ran this for a newspaper, and the utility of this should be quite apparent.
So just to answer it for all those people who are speculating why netware is still relevant - yes, it is. There are plenty of good reasons for people to keep it around. Though I'll admit it will likely become yet another good product killed by the micro$oft marketing machine.
Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.
...Chuck Norris did.
Is any of this likely to be of use to me - I have a legacy app that was written in Turbo Pascal using a file access unit that only works properly when the data files are on a Netware Server (it has close ties with NCP). We need to access the legacy data perhaps 2-3 times a year and at the moment, we just stick the caddy-based Netware disk in a spare PC fire it up for an hour or so and then close it all down - it would be ideal if there was a virtualised Netware 6.x (or even 4 or 5) that I could run on one of our Linux (or Windows) servers when required. The only other option is to port the data.
AT&ROFLMAO
I was one of several dozen people administering the Windows environment. This was from mid-98 through late 2001. We were using Windows NT workstation, NT server and I was one of the people who did the migration to Windows 2000. As of November 2000, there were still stability issues and security issues needing regular updates from Microsoft. Both of these things required rebooting the servers. I've seen several 3.x and 4.x Netware servers (from early to late 90's) that have stayed up for years at a time.
Also, it isn't 25 experienced admins, it is 25 IT personnel. I think only about 4 of them are experienced admins. The rest are techs, management, help desk, and web programmers.
The company I worked for had some of the best and brightest people I've ever worked with. The problems weren't because of lack of knowledge or skill, it was from a crappy product with scaling issues and ridiculous problems with security. The same problems many Microsoft products still suffer from today.
Since that time, I've worked in W2k environments, W2003, and I'm finally getting to work in another Netware environment. Strangely, even though we have 900 workstations and about 30 servers, we are able to provide all necessary services with only 15 IT people. Only 4 of us are experienced admins.
I've been in IT for 15 years professionally and another 10 years prior to that for recreation. I've also heard the arguments from zealots from both sides. The only ones I care about, though, are the ones from people who actually have in depth experience in both Windows and Netware. Of those people, the people who actually know what they are talking about, I don't hear a lot of praise for Windows on servers.
I've noticed a few fallacies in some of the comments that definitely need correcting. In general, people tend to combine arguments about Netware as an OS versus the services that have been bundled with it. Over the years Novell has been rather inept at developing , communicating, and executing a strategy on the inevitable migration from the Netware OS to something else. First they went through a period of simply supporting their services like eDirectory on Windows. Second, they acquired Suse and talked about parallel platforms with common services. Ultimately, it seems they made a rather smart decision in how they were going to continue to support their existing customer base that is utilizing Netware while giving a rather clear path to Linux. The problem Novell still has is that a lot of their services haven't been completely migrated to Linux yet.
1) eDirectory - Done. Has been multiplatform for years. Continues to be the single best meta directory repository on the market. There is not a single environment of any decent size that can get away with one directory to service all the business requirements, but eDirectory continues to be the best option for consolidating the directory data using Novell's Identity Manager suite of drivers and tools.
2) zenWorks - Pretty much anyone who has used it considers it the premiere tool for managing Windows clients. Only in the next release will they not require Netware for some of the components. The middle tier design and agent-based client make it a pleasure to work with compared to the fat Novell Client days.
3) Management tools - someone else already said it, but Novell cannot seem to stay focused (and enforce discipline on their own development teams) to provide a consistent management tool. They have gone from NWAdmin to ConsoleOne to iManager - except you still pretty much need each of them depending on what you are going to manage.
4) File permissions - The NSS file system is pretty damn good, has been ported and made available on Linux for a few years now. It still provides the leading access controls / inherited rights / filtered rights that other file systems should be ashamed of for not offering.
For sure, Novell is just as if not more screwed up than any other company. They have squandered many opportunities to reestablish themselves as a significant technology player, but they are hardly on the verge of going out of business. They are profitable and still growing as a company. Product lines die out and Netware has been dying out for years, but they are considerably more than Netware.
Your comments remind me that the objective "everything to everyone" very nearly defines a general purpose computer, including the entry level server. There exit better ways to implement complex systems, even though like the general purpose computer which are intended to meet, to a large degree, this somewhat nebulous objective. One architectural principle, which is very important in helping to produce a complex system, and which seems to elude Microsoft is that of "loose coupling".
Some layers or components should be cleanly separated by well designed and well documented interfaces. When loose coupling is considered to be an important design objective, you can wind up with a system in which both a rapid evolution in technology, and a stable technology and production base, are possible. (More generally, the complex system can facilitate multiple competing objectives, and let you, the client, operator, or administrator, choose at run time which objectives you seek.) The architecture permits this, for example, by providing abstractions such as "modules" which let the administrator choose what components to load and run, swapping in new modules if they need to be "on the cutting edge" or using the tried and true ones if they need stability more than any other objective, for example. As a reasonable example of this principle, consider "the web" as a loosely coupled complex system, or consider simply Apache as a very coherent example of a single system where multiple modules doing different things and created by different authors co-exist pretty peacefully over generations of software revisions and wholesale architectural changes.
When loose coupling is ignored, or subjugated to the desire to create opportunities for "vendor lock", you wind up with things like the SMB/CIFS snafu (that may sound harsh, but calling it a "protocol" is overly generous).
The items you mention in your point "2" above are nearly all rooted, ultimately, in the failure to consider the principle of loose coupling, when designing a complex system. Well, honestly as far as I can tell, Microsoft intentionally undermine it at every turn by trying to tie everything to everything else so that you get snared in the "Chinese finger trap" that is the Microsoft world.
If you mod me down, I shall become more powerful than you could possibly imagine.