Could the Cloud Derail a $300 Million Data Center?

1sockchuck writes "The cloud computing debate has come into focus for taxpayers in Washington state, where a proposed $300 million project to build a data center in Olympia for the state's IT operations is coming under scrutiny. Two legislators are urging the state to shift applications to the cloud instead, noting that two of the largest cloud computing providers (Microsoft and Amazon) are based in the state. The critics say the data center project is driven by an interest in local construction and 'fails to seriously explore the larger strategic question facing government technology today.'"

Comments

could it? Sure. Should it? No

[Em Emalb]

Good God, I can't be the only one so sick of this cloud computing bullshit. Seriously, just because it works for some types of data and/or applications, doesn't mean it'll work for everything.

Put down the fucking hammer, not every IT task is a frigging nail.

Idiots.

And here I was ...

[lbalbalba]

... stupid me, thinking that 'The Cloud', actually *was* a $300 Million Data Center...

Re:And here I was ...

[CastrTroy]

Exactly. Either you build your own data center, or you host your applications in someone else's data center. The cloud is just a term we use to describe the situation when you're running your applications in someone else's data center, on their hardware, on their system configuration, and possibly with their applications. The data center doesn't cease to exist because it's somebody elses. It just means you pay more, because they have to pay for the data center, plus get a profit from running the thing.

Re:And here I was ...

[dave562]

This presents an interesting situation without an easy, clear cut answer. As a tax payer, I'm not sure what makes more sense. Do I want the government spending my money with a private company like Microsoft or Google, or do I want them spending my money developing their own infrastructure. On one hand, it could be argued that a large corporation that faces competitive pressures in the marketplace would be forced to keep costs down. On the other hand, by having a government run datacenter, costs can be controlled through the bugetary process. If the state runs their own datacenter, they don't have to worry about their "cloud" provider raising the rates every time the contract comes due.

The politican who came out against the datacenter says that he favors transparency. It seems to me that if he wants transparency, the state should run their own datacenter because then they will be able to completely audit all of the costs associated with it. If the IT services are outsourced into the cloud, it becomes more difficult to account for exactly how the dollars are spent. As you mentioned, a private entity needs to make a profit. A public entity simply needs to cover their costs, and in fact it is quite common for legislation to contain verbage that makes it illegal for a public entity to attempt to turn a profit by charging more for services than required to cover the costs involved.

From what I know of public sector workers (my girl friend works for the state of California), they are proud of their jobs and what they do. Of course there are always antecdotes about lazy DMV workers, or life time employees who get by doing the least possible as they look forward to their pension. However by and large, most public sector work environments have a strong sense of community pride that comes from knowing that they have a job for life (budget crisises aside). They know what their jobs are and they get them done. The large majority of the delay comes from the legislatively mandated proceedures that they have to follow... the reams of paperwork that they have to fill out to do the simplest thing. The jobs aren't the best paying jobs, but they are stable.

I can almost guarantee that sense of pride would shine through with the state of Washington IT services department. That would be "their" datacenter, and they would be providing services directly to the people of the state. There will be people working in that datacenter for 20 plus years. How long do you think people stick around a typical datacenter?

Re:And here I was ...

[llamalad]

How are we supposed to have a conversation about this if you're going to go and cloud the discussion with actual facts?

Some day...

[Facegarden]

Some day, someone will figure out how to store data in *actual* clouds, and this whole thing is gonna get *really* confusing.
-Taylor

Re:*blinks*

[jtownatpunk.net]

Yeah, 'cause the various arms of government have such a great track record when it comes to network security. Remember when the department of the interior was so monumentally hosed that all internet activity was banned? They weren't even allowed to have their public websites online.

Re:could it? Sure. Should it? No

[vertinox]

Seriously, just because it works for some types of data and/or applications, doesn't mean it'll work for everything.

What is the difference between an in-house datacenter and an outsourced one?

The person you write the checks to.

That's about it these days.

Chances are if they did do in house, the techs were still be outsourced contracts instead of state employees. If they outsource it to Amazon or Microsoft in the state they'll still be employing locals and hopefully save tax dollars in the process.

But I do agree about the whole "cloud computing" being BS as a hypeword. Its really a euphemism for "outsourced".

Re:*blinks*

[Albanach]

If I lived in that state, I'd be pretty upset by the mere suggestion that it would be a good idea to have all the private information which the state holds about me go through either Microsoft or Google.

If you're hosting your own images and just using their processors, storage and bandwidth then what I would be concerned about is the privacy policy that forms part of the contract. Properly set up however, the important data should be arriving at the cloud encrypted and be stored encrypted. The host should have no ability to access raw personal data.

Personally I'm wondering just what sort of IT infrastructure they have that demands a $300 million data center? With 66,000 employees that's $4500 per employee. Just what is the data center for?

Clouds not good for old machines

[addikt10]

If this $300M Datacenter was for brand new applications and brand new servers, then they might have a point (probably not, but they might).

But in my experience, cloud computing works best for particular applications, and not as a blanket answer to wholesale moving everything in your datacenter to mystical hardware in the cloud.

I haven't seen anything in cloud computing that can handle main frame and midrange apps, Sun apps, or any of thousands of other requirements that are going to be handled in the $300M Datacenter. In the end, this is just politicians trying to seem cool.

Re:*blinks*

[eln]

Just what is the data center for?

Call me crazy, but I think it might be for storing and processing data.

This is a state government, they have data necessary to deliver government services of various kinds to the 6.5 million Washington residents. I'm not sure what the number of employees they have has to do with anything.

Personally, I'd be hesitant to build one giant data center just because you then have a single point of failure, unless their budget includes a disaster recovery site somewhere else. However, shifting personal data to "the cloud" of a for-profit company, especially when the security of cloud architecture is still being scrutinized, would scare me even more.

"Cloud" Privacy?

[mosodede]

I really do not feel comfortable with the idea of the government "outsourcing" my data to a third party. I think that cloud-computing is such a young concept that it should not be used for government purposes until any privacy concerns are addressed.

Re:could it? Sure. Should it? No

[snspdaarf]

It would be an interesting archaeology study to dig through messageboards and bulletin boards from the 70's and 80s. I'm sure that you could find people discussing the idea of shifting computing from the big, time shared mainframes to personal computers.

I'm also sure one would find comments like yours, stating how annoying the idea of personal computers sounds like the ubiquitous nail for the universal hammer problem. I wouldn't be surprised if there were several people in the vocal minority who had disdain for non-distributed computing.

Hopefully, in two decades, someone will be digging through Slashdot and laughing about us having to search through the slow-ass "Information Superhighway" for our data.

It's funny how history repeats itself.

One could. There was. And it isn't.

There is no difference between what was happening then, and what is happening now. Then, it was short-sighted management that wanted to avoid the costs of the mainframe and having to deal with Data Processing when they wanted something. Now, it is short-sighted management that wants to avoid the cost of in-house servers and desktop computers and having to deal with IT when they want something.

Then, the problems cropped up when data that used to be in one location was suddenly on every PC in the organization and out of sync with the mainframe and every other PC. Now, the problem will crop up when we will have the data on NONE of the computers in an organization, and some dork with backhoe whose parents never bought him Tonka toys chops the fiber. Or, that a poorly written application stores critical data in the clear and suddenly a Google search brings up your medical history.

When I call a company for service I do not want to be told sorry, we can't help you until whatever problem happened is fixed, because we have no way to pull up your records.

Re:Want real government savings?

[mosodede]

Windows is not quite as crappy in large enterprises as you might like to believe.

Re:could it? Sure. Should it? No

[dgatwood]

If we limit it to things the government would do, yeah, they probably could standardize on something with a dozen other states and save a lot of money. If they go into it alone, it's just a money pit, though. If you have to run a custom app just for you, I'd expect it to be much cheaper to maintain systems in-house than to contract it out to a for-profit company to maintain them for you.

That said, I can think of a lot of apps that can't realistically work via the cloud---not because of the number of records, but because of the size of each individual record and the performance requirements for accessing those records.

Examples of apps that cannot realistically exist in the cloud:

• video editing--the bandwidth requirements are too great, and each user would need the equivalent of several full blown CPUs all at once for the majority of the time (minimal idle time), so there is no advantage to cloud hosting over having those CPUs locally.
• audio editing--in addition to all the problems of video editing, audio work tends to have a live recording component. There's no possibility of using a remote computer to apply real-time effects to the audio that is being recorded. The latency alone makes this difficult even if the cloud is really a server in the next room. So at minimum, you must have enough local CPU to handle all channel effects for any channels with record activated, plus any busses that those channels dump into, plus any master effects. In short, you have to have a beefy system locally, so why bother with the cloud at all? Oh, and a packet getting dropped would spell disaster, making it decidedly not fault tolerant.
• photo editing--the bandwidth requirements are far too great for any serious editing (not talking about the fairly lightweight editors out there now), and the CPU requirements would at best allow maybe a 4:1 ratio between users and hardware, which doesn't save enough to cover the extra cost of rack-mount servers and faster networking even within your own building, much less the R&D costs or the cost of contracting it out to a third party to maintain at a profit. This ratio won't likely improve much because as computers get faster, people take advantage of that to do more complex effects and work with larger, higher-resolution images. And if you are just using Canvas on your local machine or similar, then there's no advantage to the cloud.

I'm sure there are plenty of others.

Re:could it? Sure. Should it? No

[Em Emalb]

Call me old-fashioned, but I prefer to keep my data in-house. I know who is accessing it, who even has access to it, who has access to the physical rooms, the racks, etc. Do you think that "cloud" computing is going to increase or decrease security leaks? In a few years, if this keeps going the way we're going, there will be so many leaks we won't even report them anymore.

Yeah, I do networking/security for a living, and I simply cannot trust a third party to be as responsible as I am with my data. I work for a financial firm, there is no way in hell we'd even consider out-sourcing our data or servers to a third party. Way way too much information available there, it would be too tempting, IMO.

With regards to the data center in question, yes, chances are it will be out-sourced. But I wasn't referring to that. I was referring to calling out-sourcing your servers and whatnot to a 3rd party "cloud computing". Cloud computing to me indicates that your information is distributed all over the internet, not a single provider. That's not a "cloud", that's a single entity.

But whatever floats their boat.

Re:*blinks*

[Nethead]

Those Washington State computers keep me from having to interact with State employees. wa.gov has been a leader in getting government online. I remember back in '95 when the head data guy, Jim Culp, raised a big stink with the powers-that-be by listing ALL government phone numbers on the gopher and www site, including the Governer's cell-phone. His reasoning was that the people of the State paid for it so they were entitled to the information.

I haven't been to a DMV office in over a decade because I can renew my VOL and my car tabs online. I can pay my property tax, register to vote, look up laws, ask staff questions and perform many other actions, all sitting on my fat citizen ass. All because Washington State has put so much info online. I even setup a S-corp my self just by filling out forms online.

Take a look at ftp.wsdot.wa.gov sometime. They have had this open since '95 so tax payers can dig into where their money is going.

hey folks, I'm just sayin'.....

[Rep. Reuven Carlyle]

Ok, I'm not pretending the 'cloud' is the sole answer or even that there aren't legitimate privacy/security concerns about citizen data; I'm just making the case that: 1. the state's business case dismisses the entire cloud without any real analysis of where it may or may not be appropriate; 2. the $1,200 per sq/foot cost for state data center is probably closer to $2,000 in reality; 3. the $300M is for the data center 'shell' and doesn't include anything to fix old Cobalt databases, buy or build new applications, create an XML or other neutral interface to access trapped data, get our hands on real enterprise applications; 4. even if we want to have a state owned and operated facility, does it make sense to build in Olympia where labor, energy, building, etc. costs are all higher than in Eastern Washington? 5. if we had a fraction of the $300M to buy and build very cool customer oriented applications and services, we'd be the top state in the nation for on line service, but this data center is taking every penny we have for capital infrastructure for technology. Period. 6. as for comments above about audits--I'm all for it. I agree there are legitimate issues (privacy, security, audits, etc.) but come on, at least we should design a 21st Century strategy for a new approach to services instead of automatically assuming the only answer is a state owned and operated facility. I acknowledge that I'm more interested in front end citizen/consumer applications than back end infrastructure....but at least I admit it. I don't have all the answers, of course, just trying to raise some questions before $300M of the public's money is spent. Reuven Carlyle State Representative Washington State House of Representatives--36th District www.reuvencarlyle36.com

Doing what you do best is best.

[mcrbids]

California spent a pile of money to develop their own data center in the Department of Education. A ton. And the result is less than impressive, with uptimes that approach 95%, and constant notices of downtime, often unexpected, due not only to the occasional software glitch (which happens, even to Google) but also to network issues. (Routers going down, unstable/bad performing connections, etc.)

Given the amount of money spent, the result is just... disappointing. And yet, just a few miles away, there are private hosting facilities with many times the capacity necessary for the state to host all their stuff with 5-nines uptime, with excellent performance levels, demonstrated over 5 years, at rock-bottom prices. Seriously, it isn't until you get to the "enterprise level" hosting that you discover just how *cheap* top-notch hosting is - it's a perverse, inverted marketplace, where the better the quality, the lower the price.

The State of California could have probably saved anywhere from ten to a hundred million dollars by simply renting the (highly qualified!) IT services of a local facility in Sacramento rather than trying to do it in-house. And this is the lesson that I've taken from this and many other situations: Focus on your core competence. Find out what you do best, and do that, because that's best, and outsource anything else you can to save money.

This is where Dell is about to be marginalized, because what they did best is produce decent quality machines cheap and fast, and they've been outsourcing their core competence, meaning that they no longer produce decent quality machines cheap and fast - for many of their lines, they are nothing more than a sticker on the machine. This will work for a while, but once you give away what you do best, you are a leach on the marketplace and eventually, you'll get cut out.

Doing what you do best is best.

The REAL question is: Why in Western Washingtom?

[drainbramage]

Pulling a straw man question out of their collective arse helps hide how bad a decision this is. Anyone see anything about large data centers being built in EASTERN Washington because of the very CHEAP electricity, lower property costs, and lower wages? I'm just amazed they aren't proposing this as another lid over I-5 further traumatizing traffic flow.

sometimes public policy matters....

[Rep. Reuven Carlyle]

Do you really think I'd take a stand like this for Microsoft or Google or any other company? I'm trying to have a more serious discussion before we spend $300M on a state owned and operated data center with a weak business case behind it on the Capitol Campus when we're so broke we're closing group foster homes. And I'm not suggesting we send all the data to the cloud, just look at a more strategic technology plan that uses it when/if appropriate. Doesn't take away or discount the legitimate privacy/security issues to raise other options. Reuven Carlyle