News Stay informed about the latest enterprise technology news and product updates.

Rackspace customer downplays cloud outage

Sachin Agarwal, co-founder of the blogging service Posterous, says that despite a recent Rackspace outage, his company is pleased it moved from Amazon Simple Storage Service to Rackspace Cloud Files.

During routine maintenance on a UPS system, Rackspace suffered an outage at its Dallas-Fort Worth data center this month. But at least one customer, blogging service provider Posterous, says it was no big deal. We sat down with the co-founder, Sachin Agarwal, to discuss the outage.

Describe your company in less than a minute.
Sachin Agarwal: We are a blogging service. We let users update their blogs by sending text, pictures, video or whatever content they have in an email to, and we upload it to their blog. We currently get 5.5 million unique vistors on the site per month, and we are growing like mad. We take all the hassle and work out of blogging.

How many customers do you have?
Agarwal: Can't say.

More on Rackspace:
Rackspace opens 'community store' for cloud tools

Rackspace pitches pricey private cloud

Rackspace attempts to retool hosting into cloud

How did the recent Rackspace outage in Dallas affect your business?
Agarwal: We were down for about 15 to 20 minutes; it affected a small percentage of our users. We jumped online, instantly talked to someone at Rackspace to find out what was going on and they told us. They are very transparent.

How many servers are you running in the Rackspace cloud?
Agarwal: About 20.

How much storage?
Agarwal: Twenty terabytes, but it's on Amazon S3 [Simple Storage Service] right now. We just signed up for Rackspace Cloud Files storage; we're switching from S3.

Why the switch?
Agarwal: We started the company in May 2008 using Rackspace Slicehost for computing and Amazon S3 for storage. At the time, Amazon was the gold standard, and we were not familiar with Cloud Files; it was an education issue.

We are the ones that take the hit [for an outage], not the cloud provider.
Sachin Agarwal, co-founder of Posterous,
Why move now?
Agarwal: The level of support from Rackspace; there is a human being at the Rackspace cloud who will talk to me at any moment about how all this works. [Amazon] was like a sleazy car dealer with all these hidden charges, and then last summer S3 went down for an entire day -- one month after we launched. It was bad, but we dealt with it. But if we went down for a day today, it would be massive and negatively affect our business in the long run. And it all comes down to us: We are the ones that take the hit, not the cloud provider.

What happened during the S3 outage?
Agarwal: S3 would get slow at random periods of the day, so we'd go to the Amazon blog page and it would be fully green; they were not acknowledging it. We signed up for their support services because we were so stuck. They could have charged us $20,000 and we would have said OK. But by the time we got on support with Amazon, the problem fixed itself. But we got no reassurance from them that it wouldn't happen again.

What does Amazon charge for support?
Agarwal: It costs $400 or 20% of your bill, whichever is greatest. We paid $400. But that doesn't scale. What happens when our Amazon bill is thousands of dollars and our storage costs are going up and up? We don't subscribe to a support package with Rackspace…It is part of the base price.

So price was a factor in the decision to switch?
Agarwal: No. We just wanted better support, but it turns out that Amazon charges for all these little things so we might end up saving money. For example, with Cloud Files we don't pay for internal traffic as we move data between servers. Amazon will charge you for every bit of traffic they move even around their internal network.

Have you started migrating data to Cloud Files?
Agarwal: Not yet. All new data will go to Cloud Files and then we'll start moving old data off S3.

Did the outage at Rackspace make you think twice about the decision to switch?
Agarwal: Not at all. It barely hit our radar. And there is no such thing as no downtime; it's how it's managed. Rackspace has such better Quality of Service (QoS) that it doesn't worry me anymore.

Do you think it should be possible to have nondisruptive maintenance and nondisruptive upgrades in the cloud?
Agarwal: I don't see why not. We know when maintenance is happening. Everything can be redundant and pointed somewhere else. Often, downtime is our own fault. Recently we corrupted our own database and it took us five minutes to switch to another box on Rackspace, which was a good test for us.

Would you pay more for premium support?
Agarwal: We're definitely in favor of paying more for better service. We lean towards better QoS. We have an Apple mentality. We expect quality. You pay a premium for better quality, and that's what we want. And that's not what you get from Amazon. But in general it's so cheap to run a Web site these days, there's no point in skimping when you're talking about saving $5 here or there.

Jo Maitland is the executive editor of Write to her at

Dig Deeper on High availability and disaster recovery

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.