Slashdot Mirror
The Twighlight of Small In-House Data Centers
dcblogs writes "Virtualization, cloud services and software-as-a-service (SaaS) is making it much easier to shift IT infrastructure operations to service providers, and that is exactly what many users are doing. Of the new data center space being built in the U.S., service providers accounted for about 13% of it last year, but by 2017 they will be responsible for more than 30% of this new space, says IDC. 'We are definitely seeing a trend away from in-house data centers toward external data centers, external provisioning,' said Gartner analyst Jon Hardcastle. Among those planning for a transition is the University of Kentucky's CIO, who wants to reduce his data center footprint by half to two thirds. He expects in three to five years service provider pricing models 'will be very attractive to us and allow us to take most of our computing off of our data center.' IT managers says a big reason for the shift is IT pros don't want to work in data centers at small-to-mid size firms that can't offer them a career path. Hank Seader, managing principal of the Uptime Institute, said that it takes a 'certain set of legacy skills, a certain commitment to the less than glorious career fields to make data centers work, and it's hard to find people to do it.'"
Twighlight.
"Twighlight"? Editors don't even read the titles anymore, it seems.
People pitched the timesharing computing model for a lot of reasons, lack of control of the hardware and the software rental treadmill being two of the largest. Every time I hear someone gushing over The Cloud and Software As A Service, it's history repeating itself.
Dog is my co-pilot.
In general this is true of any industry. As certain services no longer become differentiating and become commoditized, you're going to get a situation where its best to outsource these activities to the player who can do it for you cheapest.
The biggest mistake that companies make is when these data centers are part of your core business. For a University this is not the case - their core business is research and education. For my company, however, we will continue to run our own geographically redundant datacenters because they power our core business - we're a text message gateway.
That there's a 'twilight' is just the natural progression in any industry - however just figure that the jobs that remain in data center work will be directly involved with the core business. If you're at RackSpace or Amazon, then the data is your core business. If you're like us, then the datacenter is so critical to core business that you're de-facto in a position of power in the company.
Good luck to those mediocre data center managers at centers not involved in core-business. I'd start looking for a new job now.
I have a threefold problem with "cloud" storage/computing 1) Lack of control/physical security, up to and including removing my access to my own data due to a violation of some cockamamie TOS or similar agreement. 2) No ability to remedy downtimes, while rare, still do happen. 3) The ability of government agencies to scan my data for whatever they feel like arbitrarily and possibly without due process.
I hate sigs.
What level of risk you're willing to bet on your Internet connection(s).
It will be less than optimal when 20,000 kids in the school are streaming netflix in their dorm rooms while
their professors are trying to work on their research grants on the file servers in the clouds.
Combine that with multiple legal implications of the data being contained on the low bidder data center most of these kind of people will pick
From the article:
>>IT managers says a big reason for the shift is IT pros don't want to work in data centers at small-to-mid size firms that can't offer them a career path. Hank >>Seader, managing principal of the Uptime Institute, said that it takes a 'certain set of legacy skills, a certain commitment to the less than glorious career fields >>to make data centers work, and it's hard to find people to do it.'"
Which to me means "The real reason we can't find anyone to work in our data centers or provide any career path is that we're unwilling to pay anything above minimum wage."
"much CHEAPER to shift IT infrastructure operations to service providers"
It's not about easier. It's trading control, stability, and uptime for Lower IT operation costs. Executives dont care about safety of data, stability, uptime or control. All they care about is how good does the next quarter look to the board. Who cares if the company tanks in 5 years, Next quarter is all that is important.
Do not look at laser with remaining good eye.
Some people just are not inclined to trust a service provider with holding their private data, even when it is encrypted. An OEM could probably make some money by designing redundant and power-efficient storage nodes for home / small business networks to meet that demand.
Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.
That's even worse. Sales guys used to be good at producing "junk that looked pretty". You know, Pointy Headed Power Point Presentations and all that. So I guess the quality of sales isn't even any good any more.
What about McCormick?
So, in essence, they don't want to pay IT staff what they're worth and can't find enough suckers willing to be underpaid, and believe the salesman when he says his company can do all of that messy IT work for you, dirt cheap. Heard that same song sung before - remember how everyone was going to lose their IT jobs to Indian outsourcing? How'd that turn out?
Virtualization, cloud services and SAS won't replace conventional services until our 'broadband' is faster and more reliable than that offered to the consumer at the moment ...
AccountKiller
"All of this has happened before and will happen again... every five years".
You have to admire the creativity of giving the exact same concept a different name every time.
Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
IT managers says a big reason for the shift is IT pros don't want to work in data centers at small-to-mid size firms that can't offer them a career path. Hank Seader, managing principal of the Uptime Institute, said that it takes a 'certain set of legacy skills, a certain commitment to the less than glorious career fields to make data centers work, and it's hard to find people to do it.
Bullshit! There is no shortage of IT works willing and able to fill these precise jobs. But, with the promise of reduced cost and cloudy buzz words, these "data centers" are being pulled out from under these workers' feet.
Hey mister CEO, we can host your Exchange server and eliminate the cost of hardware, software, electricity, and support staff(!). All this for only $10 per month per mailbox. Never mind that over three years that's two or three times the cost of an in-house Exchange server, less staffing which could be handled by an on demand consulting firm. Never mind that when the internet blinks the whole company is down and can't even schedule a meeting. It's cloudy so, it's got to be good.
My experience in pricing these things out is that it's cheaper in-house. I can spin up a virtual machine on our VMWare/UCS infrastructure for about 1/5th the cost of a higher tier provider. I hear a lot about scalability, but so far I've never been in a position of telling somebody "I don't have room to create another VM for you." Flexibility is a semi-valid argument. It depends on what flexibility you want. If you don't need your test servers backed up, you're either paying for separate tiers in the cloud, or you're just paying for something you don't need or use. If I don't need to back up a VM in my own data center, I get direct savings from not doing so. The backups are just one example.
Cloud makes sense as an offering from 3rd party ISVs. If they have a product, they should offer a cloud option for it, where you pay them and they contract to whatever cloud provider they wish and include it as part of your cost. It's just another one of those tools that we will all use the wrong way because we have to satisfy some kind of managerial mandate. And we won't use it the right way because it jacks up the apparent cost of the products that could truly be a good fit.
Hot Damn! It's the Soggy Bottom Boys!
Every time there's an article about a shortage of workers or skills in a particular field there's some retard bemoaning the lack of workers. The labor market is subject to the laws of supply and demand just like every other market. What the whiners mean is there is a lack of workers at that particular price point. If they paid more, offered better benefits, offered more flexible schedules, or some combination of the three they'd see an increase in the supply of qualified labor. It's just that they value keeping their wages low, their benefits low, and maintaining inflexible hours is more important to them that attracting qualified workers. That's not an opinion, it's a fact manifested in the choices they're making.
Heck, I got all excited over the title at the idea of renting out an unused guest room as a data center,
I think offsite data centers will become more popular. But, I do not expect any major shift within that timeframe.
Internet access has to become much more reliable, fast, and secure, before such moves become practical for many companies.
Most of the cloud offerings don't want you using your own hypervisor anyhow. At least not without paying so much more it isn't worth it. They are looking at spinning up the servers for you, or giving you the raw resources for a linux/Windows box. If you want to implement VMware or need to have access for something like Provisioning Services, you are most likely better off running that in your local datacenter right now.
I will shred my adversaries. Pull their eyes out just enough to turn them towards their mewing, mutilated faces. Illyria
"IT managers says a big reason for the shift is IT pros don't want to work in data centers at small-to-mid size firms" that is complete and utter bull!
Now if they wanted to pay a reasonable rate then yes, maybe they'd get people to work for them. But until such time those small to medium shops stop being so cheap about what they pay their I.T. people they need to STFU right now.
This is why Bill Hicks plead with marketers to kill themselves.
The world's burning. Moped Jesus spotted on I50. Details at 11.
You're thinking of the MARKETING guys...The sales guys are the ones who tell you every feature you just asked for are in the new version being released next week then send a frantic email to the developers group telling them what needs to be added by next week.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
They're never going to do it themselves.
He'll have to do it for us.
It is the same old stuff in new wrappers. But therein is a statement:
1) In house development and app hosting weren't working. Why? Costs? Pains? Staffing? Budget
2) It's just about as secure to do things internally as on external hosting because if you do the job right, there's truly no secure boundary and people learned that.
3) Vertical market software is getting really good, and SaaS can even be satisfactory for some-- and vastly less than doing it in-house.
4) Less Capex. No huge front-end expense to setup shop/branches. Rent everything.... every IT cost is OpEx rather than CapEx.
5) Stuff moves to quickly to keep up, perhaps. Tough even for us old sages.
---- Teach Peace. It's Cheaper Than War.
They'll do anything !! And work for peanuts !! We need MORE H-1Bs !!
I don't know about you guys but gigabit is starting to look slow on a local network so you can forget about going back to 10Mb/s if the stuff is stored offsite.
What goes around, comes around. IT is and always will be the same circle continually reverting back to what it was 20 years ago.
Amazingly, the "twighlight" of data centers coincides with the twilight of spell-checking.
People pitched the timesharing computing model for a lot of reasons
And maybe the "PC on every desk" paradigm has it's own problems. But then, Ken Olsen is my hero, so I have my own biases...
"I don't know, therefore Aliens" Wafflebox1
A friend of mine's department was sent a lot of spam this week.
So gmail banned THEIR accounts for 24 hours.
I'm not sure of the logic behind that one.
She was like chocolate when she drank... semi-sweet at first and then increasingly bitter.
We've been using Saas and cloud services for years now... and it's a mess. Contract negotiations are such a nightmare with these companies, we end up employing more people specializing in "contracts" than we would have if we just kept the service in house. We recently had a major project held up for 4 months because we found out the vendor had a different "understanding" of how our data was supposed to be encrypted and they had to haggle all that nonsense out before we could move forward. Don't even get me started on Oracle...
Then you have the whole problem of: You have no control over the vendors financial well being. Not only that, but it's in their best interest to hide financial troubles from you. So suddenly they go belly up and your entire service vanishes. We had a vendor maintaining our series of websites for us and they vanished overnight. Their staff walked out, but lucky for us the owner was a reasonable guy and did his best to get all the data he could to our guys. Meanwhile we had no staff that was in the business of doing web development, though some had a pretty good idea of what to do. But once we got the data we could from the owner, it ended up parts of it were compiled and there was no source code. (I'm sure it was somewhere but the owner wasn't a developer so...) It was a freaking mess. We ended up having to run a website for months with no idea what the source code looked like for some of the more complex bits until we were able to rebuild it from scratch ourselves.
they don't teach that in CS theory based classes
The biggest drawback to out-sourced infrastructure is vendor stability and business direction.
As your vendor competes and/or refedines it's business model you can have unwelcome change thrust upon you in timelines that are inconvenient and require signifcant internal resources to manage.
Been there. More than once.
Anyone who has managed a data center understands the constant nature of changing software and hardware requirements. That doesn't go away because you've moved the problem to a vendor and you have added the risks associated with the vendor's business (mergers, acquisitions, bankruptcy, fraud, etc.) to your own.
Don't get me wrong, outsourced infrastructure can work but too many people underestimate the cost of this "cheaper" alternative by conveniently overlooking the significant operational, financial and legal risks that are aborbed as part of the client/vendor relationship.
Trends be damned: some data should not be trusted to other companies encrypted or otherwise. For example the captcha I just had is "breeches"
"he's doing it."
I know this is Slashdot and all, but can't you at least fantasize that it might be a "she" .
What your missing is that almost everything in IT has been history repeating itself for 20 years or so now. Just about everything cool, sexy, and new has been stuff the mainframe guys did long ago. Timesharing is just the latest. That's not a good thing nor a bad thing, it's just a predictable thing.
Really, the only new ideas have been AJAX (without which the web looks very much like mainframe-terminal interaction) and streaming (which changes how you want to cache stuff in an interesting way).
The industry will circulate between "centralize everything" and "decentralize everything" on a 20-year-ish cycle forever - it's just long enough for CIOs to seem clever as most have forgotten the last time we did it. In another 10 years I expect to see /. posts about "Every time I hear someone gushing over , it's history repeating itself."
Socialism: a lie told by totalitarians and believed by fools.
That's "Every time I hear someone gushing over [whatever P2P is called now] , it's history repeating itself." Freaking Slashcode will be exactly the same in 20 years, that's for sure.
Socialism: a lie told by totalitarians and believed by fools.
Hank Seader, managing principal of the Uptime Institute, said that it takes a 'certain set of legacy skills, a certain commitment to the less than glorious career fields to make data centers work, and it's hard to find people to do it.'"
I guess that makes my skills more valuable since there's a shortage of skilled admins.
There is the old saying that if you have to change an end user's mouse you are not a system administrator. Doesn't matter how many servers you actually admin. If you deal with end users you are help desk and/or a technician and can only be expected to be paid at that level.
People seem more willing to pay for depth of knowledge but not willing to pay for breadth of knowledge. So they end up with "experts" who say, "That is NOT my job". I am a professional. I get paid to get the job done. I do not care if it is Windows, Linux, hardware or software. Oh, and yes I do consider myself a Systems Administrator, even if I hand an end user a mouse now and then when the dedicated techs/help desk are busy.
I only look human.
My mother is a halfling and my dad is an ogre, so that makes me an Ogreling
People pitched the timesharing computing model for a lot of reasons, lack of control of the hardware and the software rental treadmill being two of the largest. Every time I hear someone gushing over The Cloud and Software As A Service, it's history repeating itself.
Just because your data center isnt' in-house, doesn't automatically mean "cloud" and "SaaS." I work for a company that does IT outsourcing, and what you're really time-sharing in most cases is the technical talent and expensive data center infrastructure, not the networks and servers.
Many of our customers may only have enough servers for a rack or three. They have the option of setting up redundant power, generators, weather-proof buildings, security, etc., but that would be prohibitively expensive, so we move them here, and they use ours. Even more silly is hiring several each (to provide 24/7 coverage) of firewall experts, windows experts, linux experts, exchange experts, industrial electricians, etc. when you really only need them for a few hours a month. Sure, you could try to find one or two people that could do all of that, but they tend to have to rely heavily on vendor support when they run into gnarly problems, so you're right back to having to rely on a third party.
We have a team of about 40 people that do nothing but firewalls, for example. Our customers just pay for whatever of our time they need. Basically the model allows a company to hire 1/8 of a rock-solid ASA admin, 1/10 of an exchange expert, 1/5 of an oracle dba, and half of a linux admin. Cheaper than hiring one person, and with much deeper knowledge of their fields than a jack-of-all-trades.
I've wondered how feasible it would be for me to purchase a few good servers, upgrade my home internet to business class, and provide some hosting services as a on the side, part time business. On the surface, it doesn't seem like a tremendous amount of effort or investment, other than to get a customer base installed.
I've always said English was my second language. Had Romeo and Juliet been written in C, I might have understood it.
But we still need someone on staff who knows the systems inside and out.
But turnaround on local support is sooo much faster...if only we had a few more bodies to help out the resident expert
But we could halve the time on change reqs with a dev or two
But the grass is always fucking greener on the other side.
Do you want some cheese that whine?
---- Teach Peace. It's Cheaper Than War.
He can't! He's dead, Jim!
Anytime I see Gartner attached to a study, I am fairly certain the conclusion is utter garbage. PHBs eat that crap up and ask for a second helping.
That's hilarious, but 100% true, the main allure of the cloud right now is infrastructure automation... but what's that done through?
Cloud OS / software.
Say due to future outages the cloud providers aren't found reliable anymore, that cloud OS / software will go right back in-house to the businesses in a thousand different flavors again.
It does tend to get a bit better each time though as hardware advances and software follows suite.
People pitched the timesharing computing model for a lot of reasons, lack of control of the hardware and the software rental treadmill being two of the largest. Every time I hear someone gushing over The Cloud and Software As A Service, it's history repeating itself.
The possibility of eliminataing a lot of salaries and the corresponding bonuses for the suits are making them really push hard for cloud and SAS this time.
I still haven't received a good answer for when the whole thing falls over one day and people do not have access to their data. Or maybe even their Applications. And it will happen.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
That's hilarious, but 100% true, the main allure of the cloud right now is infrastructure automation... but what's that done through?
Cloud OS / software.
Just wait until the suits find out they cannot throttle the IT guy in order to get something done, then find out that to their holy cloud provider, they are just another customer.
It's the same thing with insourcing versus outsourcing, which is the twin brother of the cloud. Just try putting heat on the outsource folks to shorten their deadline. They might have alot of customers, and putting you first puts others later, and you aren't worth the trouble. You can save money outsourcing, except when you can't.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
How right you are! But for longer than 20 years...consider IBM's DISSOS, circa 1980.
Oh, cool, that's new to me - and I'd never seen the SNADS protocol before.
Socialism: a lie told by totalitarians and believed by fools.
Only if it's authentic RMS Cheese.
Cloud utilization is growing. And it's growing in startups and small companies. The reason isn't because of career choices by IT professionals. It's because it's a lot easier to buy a cloud-based solution with your company credit card than to requisition a VMware cluster.
Much of Amazon's cloud customer usage is for shadow-IT and small startups who do development work. Microsoft spent over $3bn on Azure and has little to show for it. Of course, object storage is a no-brainer for streaming content providers because who cares where you store a large block of data.
Regarding uptime and connectivity, Amazon suffered a major glitch last year that tanked Netflix for about a day because they didn't have enough connection redundancy. Their are providers out there who do. One I know of has multiple availability zones in the US, 3+-homed internet, and power from at least two non-connected grids.
Organizations are moving to the cloud, but large enterprises are not moving their legacy applications to the cloud. Yet. It's really hard to migrate 1000 applications running on legacy hardware, some of it with outdated OS's and non-x86 hardware.
It will eventually happen because companies are sick of having Chief Electricity Officers.
Totally believable that Universities have been moving their systems off campus. Our local State University uses gmail, google apps, and a Cisco phone network provisioned and run by a local telco. This I believe. However, the wheel fall off the wagon with some some dumb statements like, "IT managers says a big reason for the shift is IT pros don't want to work in data centers at small-to-mid size firms that can't offer them a career path."
Yeah, I want to meet that manager.
Whatever. The career path is dermined by the industry's needs, not what Mitch the IT Guy wants. I can imagine the "IT Manager" who made that statement for the article.
Moving in-house routine IT support to an outside entity is an extremely good way of doing things. Your entire infrastructure can consist of chromebooks, with a few network attached storage units sprinkled throughout the office. Your personal private secure data is stored locally, as well as in the cloud. The applications are in the cloud. If the cloud breaks, you still have your data stored locally. If your nas should break, you still have the cloud as a redundant backup. A single visit per year to set things up and look at needs. You can still host highly specialized applications on a local server if you want, but for most routine office tasks (spreadsheet, word processor, email, appointment scheduling), the cloud is your baby. Since connecting to the cloud can be done anywhere, your local NAS needs to be able to sync with the cloud on a daily basis. Licensing rates of a few dollars per seat per year and there is no reason in the world why anyone would want to do anything else. Better yet, adopt ODF format, and you aren't locked into any one vendor. Your data is mobile, and you can cherry pick license rates. The obvious upside to all of this for IT people is working for an IT organization, and not an 'accounting office' or 'diaper delivery service' or other company who 'needs computer something' but doesn't quite know what you do, offers no compensation outside of what they read somewhere about the going rate, doesn't know about system complexity, and doesn't care. Working for people who know means appreciation and understanding.
Well... managing infrastructure in the cloud = managing virtualized infrastructure... so it's just like VMWare w a GUI that puts VMWare to shame (the GUI is the cloud OS). So there's still an IT guy that logs into the cloud and sets up the extra sever or db or w/e... and no way is that free as with a private VM stack (if you have the server licensing, you need the licensing either way). So... as your infrastructure grows, your cloud costs grow, but that's almost soley offset by no need to upgrade the hardware.
While I can spend all day making fun of the suits and how out of touch all of them are with IT (especially the ones who think they're in touch w/o writing a single line of code), that's not really a problem in the cloud.
Where the cloud provider would tell the suits to go f' themselves is on cloud provider downtime, AWS's response every time has been "We're working on it, no eta". Which is enough for a top exec to grow some new grey hair.
Here in healthcare we'd love to be able to use more "cloud" services, where it makes sense. One of the main problems we have is a stated lack of HIPAA compliance. Also the ability to integrate these services. For example, how do I integrate my "cloud based" (read: web app) health information system with my web based office suite and do it securely?
There are a few places where the cloud is great. If your machine is lightly loaded and it is doubtful that anyone really wants to steal your stuff. Then the cloud is great. If your load is all over the place but averages fairly low then again the flexibility of the cloud is great. If your customers are all over the place and distributing your stuff here and there is good then again the cloud is great. If you are risk taking fast moving startup then the cloud is for you.
But if security is off the scale critical (you are a bank, phone company, hedge fund, etc) then you must have physical control over your machines. Also if your machine uses a fairly steady load then your cloud costs will soon dwarf a dedicated machine. Then there are many other cases where the cloud is just stupid. If the main users are say within one building and any connectivity would stop most work then remote machines could pose a huge risk.
Then there is the third option which is you run your own cloud but control the machines physically.
So I would say that the future of the cloud is (to use a bad pun) murky. It is a great option for some a neutral option for others and for a few a terrible option. But where so many people have so many different needs catch all statements about the future of the cloud are nearly useless.
As one commenter said this is the same old crap new wrapped in todays newspaper. I just feel sorry for the IT people who work for some company where the MBA types are all "We've got to join the cloud before we get left behind."
You make a good point on staffing. I work for a SaaS provider and we have good sized footprints at redundant data centers. Our clients pay us to host their applications because it is less expensive for them and we can leverage economies of scale. There are also regulatory considerations that make it cost prohibitive to host certain kinds of data in house.
The biggest challenge however is staffing. There are simply not enough competent IT people in the world. The need for talent outstrips the available talent pool. At the upper levels of the game, junior people cannot get the hands on experience that employers are working for. It is a chicken and egg problem. People need SAN experts, but how do people get SAN experience? The same thing goes for anything at scale, be it databases, web farms, Exchange, virtualization, whatever. Companies can hire people and mentor them and bring them up, but anyone with the skills and abilities to learn what needs to be done will also quickly realize that once they are capable of performing on their own, there is serious money to be made in consulting or working somewhere else.
Although the article discusses data center consolidation, the real trend that I have observed is talent consolidation. It is not easy to build up the kind of organization required to support 99.999% uptime. The tech is one thing, but then you have to have the processes, be it ITIL or ISO 27002 or COBIT. You need the security infrastructure. You need the regulatory infrastructure (SOX, PCI, HIPAA, whatever). You need true HA within the site and DR for when the unthinkable happens. Running an effective IT infrastructure takes a small army of people to do it right, and those people are in high demand and can command whatever salary they want. There is a lot of money to be made in the SaaS model, but there is also a lot of risk.
I still haven't received a good answer for when the whole thing falls over one day and people do not have access to their data. Or maybe even their Applications. And it will happen.
It is called an SLA. You do due diligence when you select a provider. You do an audit to make sure that they can do what the sales monkeys claim they can do. You make them conduct a full blown DR test of your application stack. You make them promise to continue to do tests on a quarterly or bi-yearly basis. You get it all down in an iron clad contract that stipulates the penalties for breaking the contract.
If they do not succeed, you do not do business with them.
When you find a provider that can do all of the above, do not be surprised when they charge you an arm and a leg. There are a lot of people who claim five nines uptime, full multi-site DR, blah blah blah. There are very few who can actually deliver it. Those who can deliver it have more prospective customers than they can handle, and therefore can demand high prices for their services.
This, and trusting my data to another company is scary as hell.
I'm poor. Please donate. http://albanypcs.com
All that means is when your data does disappear, your provider will go bankrupt right along with you.
I still haven't received a good answer for when the whole thing falls over one day and people do not have access to their data. Or maybe even their Applications. And it will happen.
It is called an SLA. You do due diligence when you select a provider. You do an audit to make sure that they can do what the sales monkeys claim they can do.
Failed answer. You are trying to impress me with Alphabet soup.
What is worse, all your non-answer is saying is "If you do not do this or that, it is your fault." Which is why I say that the Cloud never fails, we can only fail the cloud.
There truly is no true Scotsman as far as the cloud is concerned.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
I am trying to educate you. Obviously I failed.
I work in IT operations for a SaaS provider that generates over $60 million a year in revenue. We give our clients guaranteed SLAs and back them up. We host a small handful of specific applications targeted at a narrow market. We are not a traditional "cloud provider" where we let any monkey install whatever application they want. We do have a "private cloud" with excellent up time and our clients are very happy with the service we provide.
Like everything in life, you get out of your service what you put into it. If you are not willing to do your homework and due diligence, you will get bit in the ass. If you know enough about the service you are buying, you can find good providers. But expect to pay for it.
I think the "alphabet soup" confused you because you do not understand the technology. To anyone who knows anything about highly available application stacks, what I said is very clear.
Yes, but prefer X, NX, virtualGL and even VNC. I'm at home doing some stuff remotely today with x11vnc since I want to get to stuff I've left in a couple of dozen windows all over my work PC and can't be bothered to set it all up again at home. While such things work they hit limits which can become unacceptable.
If you have enough people on site trying to get stuff from a datacentre citrix/X/NX/virtualGL/turboVNC is still going to suck trying to push all those sessions down the pipe, so you may as well have everyone stay at home to avoid clogging that single pipe. Sometimes that's enough of an answer, at others (even just shifting a few hundred MB around) it's a showstopper.
Anyway my point here is that farming things out to do stuff remotely adds a bottleneck a few orders of magnitude greater than if people are on the same network as their data. There are workarounds for that limitation in some situations but in others it sucks - if the user has a few hundred GB on some media that needs to go onto the system they are going to have to wait a very very long time before it all gets somewhere at the other end of a 10MB/s link, and they are already bitching about it being slow on gigabit.
Some people can get away with the limitation, others can't. As for me, I'm currently looking at what I need to do to go 10Gb/s on chunks of the local network and to make sure stuff comes off disk at higher speeds so I'm not interested in a 10Mb/s bottleneck.
So this is all just the same as the mainframe days? Really??? I'm getting tired of hearing that, actually... almost as much as hearing the term "cloud" bandied about
I am trying to educate you. Obviously I failed
Yes, because you didn't answer my question. I'm saying that at some point, the cloud is going to be a big bust for people. It might be from an outside attack, it could even be from competitors.The closest thing I can get out of your reply is No it won't. You obviously think I'm mentally challenged, so be it. But a large part of my career has been spent by asking dumb questions of how might something fail. I can tell you that it's a great way to be unpopular in a room of excited people who believe a concept cannot fail. Only until it fails.
Your response that since I question your use of acronyms is indicative of my ignorance? Okay, no problem at all. I am woefully ignorant. That's why we have people like you who are smarter than people like me. So I have it on your authority (because you are arguing from authority) that the cloud cannot and will not fail?
Because you are in essence saying that the cloud cannot fail, only we can fail the cloud.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
Well this is "history repeating itself" with us being further along the Moores Law curve, and we know a lot more than "back in the day".
What your missing is that almost everything in IT has been history repeating itself for 20 years or so now.
Such as mobile devices for example? Social Networking? Ecommerce and Web 3? Multimedia convergence? Perhaps what you meant to say is that one small concept of IT infrastructure design is a little bit similar to how things we're done 20 years ago, but as for most of IT, it is mostly all new and different.
All hardware sucks and all software sucks. The difference that has to be weighed up is wether the applications/data are going to fail less and is going to be more recoverable if it's hosted in the cloud than if it's hosted locally. If the business has the time, the human and monetary capital and the equipment to be more reliable, with more access to the data than if it's hosted in the cloud, then there's no reason to host in the cloud. However, for the price/performance/reliability offered nowadays the cloud is hard to beat when all those factors are taken into consideration.
I understand being unpopular. I have had a similar career experiences and although it often times takes months or years for things to play out, the satisfaction and recognition that comes from having foreseen the challenges long before others did is very rewarding. Have you made it to the point in your career where you can mitigate the failure? Have you reached the point where you architect systems to be resilient to failure and continue running? It is one thing to perceive a risk. Anybody can do that. It is another thing entirely to mitigate the risk and develop solutions to address it.
I tell you that you are ignorant because you do not seem to understand the technology. You do not seem to understand the market drivers that pushing the industry towards "the cloud". So let's clear up a few things.
Technology fails. You are not some guru because you can predict that technology fails. The industry has been dealing with failure for decades. Hard drives fail so they RAID them. Servers fail so they mirror them. RAM fails so ECC was developed. Power fails during writes so battery backed cache was developed. Do you see the trend here? Virtualization is all about improving up time and reducing the impact of failure. With some of the SAN replication technologies available, entire data centers can fail, petabytes of data can go offline and then be available across the country or across the world in less than hour.
As a business model, the cloud is not going to fail because the economies of scale are making it cost prohibitive for smaller companies to do IT in house. Ten years ago, companies did not have Gmail and hosted Exchange as options for email. Now they do and only the largest corporations still run their own email infrastructure because it is too expensive for a small business to setup the highly available infrastructure required to guarantee 24x7 up time for email. For example the company I work for did over $900 million in revenue last year and we are not hosting our own Exchange because it is more cost effective to outsource it.
The same goes for large LOB applications. The SaaS model works. At scale it is less expensive for one company to run Oracle for 50 companies than it is for 50 companies to run Oracle for themselves.
Ten years from now, you are going to look back on this conversation and realize that you were on the wrong side of history. The cloud model is not going anywhere because it is built on top of all of the technology trends. It is the pinnacle of everything that the IT industry has been building for the last fifty years.
I think
should probably read:
Except for a trading group that I worked for back in the '90s, no business that I've worked for had developed (or maintained) an alternate means of running their business should their IT function fail. They were wholly dependent on having access to their IT infrastructure. (Note: wasn't there a recent /. article about the increasing stress levels experienced by IT folks? I think this is one of the reasons for that stress.) The way some companies are jumping into cloud services is scary. If they lose their cloud access they may as well send everyone home. SLA or no SLA, lose access to the data and the applications and some companies are going to fail. All an SLA will do is give the lawyers something to fight over while everyone else is cleaning out their desks.
I will agree that there are some of these outsourcing arrangements that really are as bulletproof as they claim to be and can withstand most types of outages. I strongly suspect, though, that a great many of them aren't. Yet the people who are actually running internal data centers are under constant bombardment by the marketing of the outsourcers who are schmoozing with the folks in the corner offices and come to believe that they just have to do this outsourcing thing, or move to the cloud, or adopt whatever shiny thing that was in that article in the in-flight magazine. The hell of it is that the people running the data centers that are targeted for moving to external providers are quite often the people most knowledgeable about how their applications operated. And they'll quite likely all be let go. After all, you don't need those propeller-heads who knew about such things as servers. But do you think the IT staff at the external data center provider knows -- or even cares that much about -- how your applications work together and will be able to help when you need to migrate or make extensive changes? I wouldn't count on it.
The posters who are saying that there will be failures of outsourced data centers or clouds are, I think, spot on. There will be. They may not be failure in the sense of the data center going down but there will be failures just the same. It'll just be someone else's fault when it hits the fan. That'll sure be comforting.
Just my $0.02...
CUR ALLOC 20195.....5804M
You are absolutely correct in all of your observations. That is why it is so important to vet any cloud provider, and then continue to do yearly audits.
My organization deals with large financial institutions, among others. We have auditors in our data centers and going through our IT operations on a yearly basis. We are in constant communication with the clients' risk management team.
At this point, the "cloud" is too new for everyone to just jump aboard. I would say that unless a company understands the risks involved in the cloud and has the staff to appropriately manage the risks, they should not trust their operations to a cloud provider. The companies who are a good match for cloud providers are those who have taken a good look at doing it themselves and decided not to due to the cost or complexity. Companies in that position know enough about what it takes to do it based on their own planning. They are in the position to not be fooled by smoke and mirrors from the vendor.
The cloud can work, but it takes a lot of staff and skill to do it right. Right now the margins are there to enable the early adopters to do it well. The pricing pressure has not come to the market in such force that it is impossible to do it right and be profitable. In five years from now, it is going to be a different story. The current players will have established themselves to the point where they are leveraging economies of scale to drive down costs. New entrants into the market will not be able to compete on cost.