1 Comment

Summary:

Cirtas Systems, a maker of storage appliances, has hit hard times. The company, which raised $32.5 million, is now cutting a large percentage of its workforce and going back to the drawing board after a contentious board meeting that left it minus at least one investor.

dark clouds

Cirtas Systems, a San Jose, Calif.-based maker of storage appliances, has hit rocky times. The company, which just exited stealth mode in September 2010 and had raised $32.5 million, is now cutting a large percentage of its workforce. The company, which so far, has been making gear for directing primary data to and from the cloud, is going to go back to the drawing board after a contentious board meeting that left it minus at least one investor.

All appeared to be going well for Cirtas until Friday afternoon, when Dave Graham, a senior systems engineer, tweeted he was among about 20 employees the startup was letting go. The official company line is that “Cirtas has decided to pull back from the market at this time. Through our deployments we’ve learned a significant amount about what is required by the market and believe it is in the best interest of the company, our partners, and customers to focus on what we’ve learned.” Cirtas CEO Gary Messiana told SiliconANGLE’s John Furrier, “the product needed to be retooled for extensibility” and that the company is now focused on rebuilding it to address these issues.

What Went Wrong

Details about what exactly led to this turn of events are vague, but it appears the company’s Bluejet Cloud Storage Controller, which routes primary data to cloud-based storage such as Amazon S3 and lets customers interact with that data like they’re using traditional enterprise storage systems, wasn’tt living up to customer expectations around performance and stability. It wasn’t living up the company’s expectations, either. Deadlines for fixes weren’t getting pushed regularly, leaving employees in the field with the unenviable task of explaining why promised features weren’t yet ready.

Ursheet Parikh, CEO of StorSimple, another storage-appliance startup, said Cirtas’s problem was that it committed a fatal storage mistake of putting a product into the market before it was production-ready. People don’t get fired for buying expensive storage, he explained, but they do get fired for buying faulty storage. Regardless whether it was a fair expectation, if you’re advertising a storage controller for primary data, the product has to be highly available and stable before it comes on the market, he said, or potential customers will start asking questions.

These customer concerns might have been Cirtas’s death knell. According to Parikh, acquaintances present at a Cirtas board meeting last week relayed a contentious situation that resulted in secondary investors pulling their investments because they weren’t getting what they were promised in terms of the product. At this point, Bessemer Venture Partners and Shasta Ventures have reportedly rescinded their investments in Cirtas. By all accounts, it looks like Cirtas is now trying to get its intellectual property in order for a sale so that remaining investors can get some return. CEO Messiana noted to Furrier that there have been offers to buy the company.

Did Cirtas Act Too Rashly?

What’s interesting, though, is that the product’s shortcomings didn’t appear to be as fatal as the board’s strong reaction to the lost funding would suggest. I spoke with Graham, who tweeted the news on Friday, and he explained that there were plenty of customers who were “ecstatic” with the product as-is. He referenced a couple of big proofs of concept that were underway with major customers, including one that actually kicked off on Friday. The reception was generally warm, he said, and from his perspective of being out in the field dealing with customers, the market for Cirtas’s product is warm, too.

Regarding absent features and enhancements, Graham explained that part of his job was to teach customers that they shouldn’t look at cloud storage — especially an appliance — like they do traditional storage controllers, because the two are very different means of delivering the same end. As Graham explained, you don’t necessarily expect hardware-based high availability, for example, in an appliance. The problem might have been a marketing one, and those can be remedied.

I can’t help but think that things might have turned out different if Cirtas had been bringing in significant revenue to allay its investors’ concerns about the product. But, as Graham noted, the company’s sales model involved a long proof-of-concept period, so even happy customers weren’t necessarily paying customers yet. At StorSimple, for example, Parikh noted that the sales cycle is usually between 90 and 18o days from first meeting to purchase order, and can get complicated because, as likely is the case with Cirtas, this is many customers’ first foray into cloud storage. They often want to call in the legal team and undertake other precautions.

Whatever happened, it’s further proof that cloud storage is a tough business. Cirtas was the second cloud storage business to fall last week, after Iron Mountain pulled the plug  on its overpriced and undifferentiated Virtual File Store business, essentially ceding that market to cloud-based providers such as Amazon Web Services and Nirvanix. In that case, it appears that Iron Mountain either couldn’t distinguish its cloud service from cloud-first providers in order to justify the higher price, or couldn’t achieve the economies of scale necessary to compete on their levels.

We awaiting confirmation from Bessemer Venture Partners and Shasta Ventures regarding the status of their investments in Cirtas Systems.

Image courtesy of Arjen Bax.

You’re subscribed! If you like, you can update your settings

  1. Filesystem-to-cloud storage bridges are an interesting, but ultimately short-lived product solutions. Cloud storage like S3, CloudFiles, OpenStack Object Storage will ultimately be used directly by the tools, programming languages & frameworks and end-users. Filesystem adapters are fundamentally architecturally incompatible with these types of storage products.

    Filesystems are good at transactions, random file access and can be convenient for end-users (a folder / mount on your system).

    Cloud storage systems deliver none of those well, but instead offer a convenient tool for application developers to store and deliver data to their applications and end-users. If application developers use a service provider like Amazon/Rackspace, the application developer doesn’t need to guess how much storage they’re going to use.

    If IT directors / CIOs are expecting the same suite of services of a traditional SAN/NAS system out of an object storage system (that’s remote!), then surely they will be disappointed. There is a ramp-up to adapt their applications to this new storage paradigm, rather than paper over the interfaces with seemingly-compatible filesystem emulators.

    If a do-it-yourself cloud storage system like OpenStack Object Storage is used, some of the latency issues are addressed, but it will fundamentally still behave like S3/CloudFiles equivalents. Which is to say, it will be a system that loves concurrency (with it’s horizontally-scalable storage tier), is very durable (by storing in triplicate), is very available (with a horizontally-scalable access tier) and is very cheap (as it runs on commodity gear).

Comments have been disabled for this post