You are currently browsing the monthly archive for July 2010.
I just got back from Austin, TX and the 2010 Hostingcon conference is fresh on my mind. First off, this was the best Hostingcon I’ve attended in several years. I sensed a re-invigoration in the hosting sector at this year’s event. Perhaps we are coming out of the economic recession’s malaise over the past two years. Everyone was pretty upbeat about new growth in 2010 and the coming years. I think across the board most of us in the hosting and data center industry saw flat or single-digit growth in 2009. This was certainly a shock for an industry that experienced double-digit growth for many years.
I think much of the buzz at the conference was due to the arrival of cloud computing and new related growth strategies. It seems like every 5-7 years the hosting industry reinvents itself. I remember Hostingcon five years ago in Chicago where much of the conversations evolved around cheap dedicated servers and shared webhosting.
It was clear that almost every hosting provider was building a cloud computing platform or planning to build a platform. Most providers were looking to partner with cloud platform vendors versus trying to completely build out a platform themselves.
It was my first time visiting Austin and I really liked the city — although it’s pretty darn hot there in July. We had great food and entertainment every night. I dubbed this year’s conference as “Partycon” due to all the vendor parties each night. We attended 3-4 different parties each night walking down 6th Street. I honestly don’t recall seeing so many vendor parties at one conference since the dot-com days. Fun!
I also had the opportunity to speak at this year’s Hostingcon as a co-presenter at a cloud session with Shannon Williams, VP at Cloud.com. My goal was to provide a real-world overview of what it takes to build and launch a cloud computing service. I talked about some of the challenges we encountered during our build process and the techniques and tools we used to overcome those challenges. I received really positive feedback after the session and I know the session participants had solid takeaways from the presentation.
I’m looking forward to Hostingcon 2011 next year in San Diego. That’s another city I’ve never visited but I’ve heard so much about. My only recommendation to the conference organizers is to improve the quality of the conference food. The food options at restaurants around Austin were amazing while the catered food at the conference was really poor — in diversity and taste. You gotta feed the stomach if you want to feed the mind.
I liked the Parascale guys. I had great conversations with their CEO Sajai Krishnan over the past year. Their team listened to our feedback and incorporated some of our ideas into their storage platform. Our operations team liked how their platform supported both the front-end storage access and the back-end storage management. Our CEO liked the price. Then, in June they weren’t able to secure another round of funding and the company suddenly blew up. This happened shortly after we launched our new cloud storage service on ReliaCloud using their cloud storage platform. Sigh.
A year earlier when my cloud team was reviewing different vendors and technologies I warned them that there was the distinct possibility some of these vendors would fail. My future prediction turned out to be right.
As a Parascale customer and informal product advisor I have some insights into why they didn’t succeed. I don’t know what went on behind the scenes in their business. But I’m not sure that mattered much. I think this was a case of a company not understanding who its customers were and not reacting fast enough to the needs of the market. Here are my thoughts:
- Was their target market service providers or single enterprises or both? The needs of a service provider are oftentimes different than enterprises (think multi-tenant, provisioning API, scaling, etc). When we initially engaged with them they were very interested in working with service providers but by the end they admitted they were solely focused on enterprise customers.
- The platform had scale-out problems due to NFS and the single meta DB design. NFS just doesn’t scale that well — especially with hundreds and thousands of potential connections, many originating over the WAN. WAN latencies kill NFS scaling. That’s why HTTP is the standard protocol for accessing cloud storage.
- HTTP Rest storage access was firmly fixed in their road map when we reviewed the technology in 2009. But the delivery dates kept slipping until we were told in Spring 2010 that HTTP was off their road map. That was definitely a warning signal and in hindsight we should have dropped the platform right then and there. The bottom line is that you can’t offer a cloud storage solution today without supporting HTTP storage access. I think as they began to focus solely on enterprises they saw less of a need for HTTP support. That’s probably true but then they become just another NAS solution — albeit one that has some interesting scale-out characteristics.
- It took them awhile to release their provisioning API and once they did (in early 2010) we found it to be pretty buggy. Their account management and service provisioning scheme was too complicated and simply didn’t work a certain percentage of the time. Contrast this to typical cloud storage services which are generally drop dead simple in terms of authentication and account management. Simplicity is a key characteristic of cloud storage. Developers don’t use cloud storage because it is better than other forms of storage. They use it because it is less complex to access and manipulate. You can access if from anywhere using a variety of platforms. It’s limitations are its advantages in some ways.
- Cloud ecosystems are now absolutely key and Parascale was never able to create an ecosystem of developers or 3rd-party vendors. Contrast this to a cloud storage vendor like Mezeo that has done a much better job of attracting vendor support. Ultimately it is applications, tools, and solutions which drive cloud storage adoption — not just end users.
I send my best wishes to the former Parascale guys finding new opportunities in the future.
