« Hey Now! 3PAR Parties like its 1999! | Main | Steve Duplessie on the dynamics of cloud computing »

May 14, 2009

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00e553e34fa48833011570881e93970b

Listed below are links to weblogs that reference Google's outage and real cloud computing:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

nate

My company priced out cloud computing both from Amazon and Terremark as part of our recent decision around expansion. At least for our particular purposes over a 3 year period the costs involved with Amazon was more than double than doing it ourselves. Terremark was almost 4x more than doing it ourselves. Interestingly enough Terremark was within $20k (over a 3 year period, so basically the same price) of the pricing of a managed Rackspace solution(using real HW not VMs).

We may be the exception rather than the rule in our ability(?) to operate in a very efficient manor.

I'm of course happy about the result as an infrastructure person.

Terremark seems to have the best VM solution I've seen. AT&T and SAVVIS are far behind, at least as far as being able to spin up capacity on demand, both of them(last I checked) required upwards of 5 days to get new capacity, when the ideal world you can get more capacity in minutes. Amazon offers this to some extent but their service has many other limitations in other areas as well, which didn't make it too suitable for this particular project.

The biggest failing of the Amazon service is the lack of a high performance load balancing solution. Some folks deploy open source load balancers to work around it, but those just don't offer the performance we need.

Maybe Foundry/Brocade can convince Amazon to deploy their new high end load balancers and offer their services to potential customers. The ServerIron ADX 8000 is the 3PAR T800 of the load balancing world, not only is it the fastest(1M+ connections/sec 70Gbps), but your also able to virtualize it as well(more like hard partitioning than soft virtualization).

Terremark has a very nice integrated firewall-VM-load balancer solution. Though I don't remember the load balancers they use off hand I think perhaps Netscalers, not too famillar with them.

joseph martins

Isn't this usually how it plays out?

Bleeding edge companies create solutions that do not exist commercially, because they have no choice. Eventually commercial products are launched and catch up or surpass the home-grown solutions.

Then the bleeding edge company is no longer on the edge, but instead between a rock and a hard place wondering if it should continue to invest in its homegrown solution or make the leap to one of the off-the-shelf alternatives.

I'm sure this happens quite often. The most memorable one for me was a former competitor, Vignette, that developed its own application server in the 90s. Eventually the company bit the bullet and ditched its own solution once proven viable commercial alternatives were available.

joseph martins

Nate, your situation is not at all unusual.

We did a TCO study for one of the well known "cloud" providers in the Spring of 2008. Something like a seed of 50TB with 50% y-o-y growth over 5 years. We looked at online cloud vs disk vs tape vs disk/tape hybrid.

The results were rather surprising to their MBA marketing type who collaborated with us on the project. I believe they expected to see a substantial difference over in-house solutions. In fact, in-house can be more cost effective depending on the requirements.

Their first reaction was that our numbers must be wrong. Their second reaction was that we should be using the numbers of [insert other analyst with more favorable numbers here]. Obviously we're not going to do that. And their final reaction was to refuse to pay the balance due on the project. Nothing irritates me more than vendors who can't stand the sight of their own blood.

The reality is that there are many hidden costs in the fluffy white cloud. Three of the big ones are:

1. The pain and cost of moving data into the cloud and reconnecting the dots between your data and your applications.
2. Cost savings from the use of deduplication on the backend IS NOT passed on to the customer. The impact can be very costly, especially in high availability environments with multiple copies where they charge for the capacity of each full copy.
3. The cost of bandwidth to move data to and fro depending on the size of the pipes.

So you're not alone. As for that provider, the VP mktg and the MBA moved on to pursue other opportunities shortly after that project.

sduplessie

Markster, good thread. Let's remember that Google didn't initiate it's company with any desire to build a "cloud". It was and is a search company. Amazon didn't start out to be an IT Cloud provider - it's an online retail aggregator. Both of them HAD to roll their own stuff, because traditional industry offerings simply didn't (then - nor now IMHO)have what they needed - namely infinite dynamic scale of ridiculously cheap commodity stuff to completely unknown levels - with no business model to support it and no people to manage it. IT 1.0 doesn't work like that.

Frankly, it's clear to me that EVERYONE will want these attributes - and also that traditional IT offerings are simply not built to provide them. Worse, it's impossible for any vendor to stop taking a buck from someone who is happy to give it to them - even though they know that the stuff they are selling them is 47 generations old and about as useful in the new world order as me owning and maintaining a fleet of saddles.

So, who cares if Google's service goes down? How much are you paying to use it? For the overwhelming majority of data, it's probably just fine. If you absolutely positively have to be able to count on being able to do your thing in the cloud - or in your own IT operation - then you are going to spend money. I have a lot of trouble believing that Yahoo mail or Gmail are any less reliable than running Exchange - and we pay lots of money to run Exchange (service based, or on premise).

Thus, the point is that this is NOT a zero sum game. It is a fact, aided by common sense, that unless you WORK making a living by performing IT functions, the business would rather have its cake and eat it too - they want all the access, availability, and ability to react instantly to new demands without any of the bullshit that typically comes along with it. If that means a VMware enabled cloud because IT is more comfortable with that during this transition, so be it. That does not mean that it's all or nothing.

I'm pretty sure that Google runs it's billing systems on "real" gear, with real app's, and real IT people. Ditto Amazon.

We're in the normal stage of "mine is bigger than yours" that always occurs at turbulent market inflection points - but that's not the right battle. We can't even agree on what the cloud means, let alone what the customer will part $ for yet. It will end up that the entire economic and operational model of the last 50 years of IT will get turned on its head (finally) I'm sure - but that doesn't mean that there is only room for one. Cloud is a way of life, not a product. It's a construct for performing a business function - a valid one, who's time has come - that will change things forever. (Remember, we could still ride horses to work, but that would be stupid and impractical in the current realities of life). The basic principals of cloud - scale, dynamism, cost efficiency, etc. are legit. Let's not get too hung up on product A versus service B before we agree that these are indeed required attributes. Then we can figure out the best way to implement in and out of our companies - and in between our different data/compute requirements. Show me the audit trail - prove to me that you did what you said. Prove you are secure and my competition can't steal my secrets. Live up to the hype and eventually I'll come along, dragging my checkbook with me.

The days of one size fits all are toast. --- Steve

Dave

I'm not sure Google wants to 'catch up' Mark.

nate

I thought this article was great: http://www.theregister.co.uk/2008/08/25/cloud_dziuba/

Don't know how accurate it is, but it was a very enjoyable read around the Amazon and Google cloud stuff.

marc farley

Thanks Nate, an interesting read.

kevin

As Dave has mentioned, I am not too sure Google would like to catch up. Of course by providing gmail, they are running the SaaS model. However webmail were already around way before then, and Google just provided it because they could (or possibly to irk yahoo then)

Just the temporary glitch and no one would complain to Google about the downtime, and demand compensation for a service that is supposed to be free. One starts to wonder we are getting to reliant on Google, and what standards they are setting for cloud computing.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

Comments are moderated, and will not appear until the author has approved them.