(WEB HOST INDUSTRY REVIEW) — As a non-profit organization, the Wikimedia Foundation (www.wikimedia.org) has relied on the donations of individuals and groups to maintain the popular Wikipedia (www.wikipedia.org), and its other sites.
So it’s not necessarily surprising that these sites have had issues in the past with outages, as in the case of the July 5th outage which affected all Wikimedia sites following power failures at its Tampa, Florida data center.
Currently, Wikipedia and its related sites are supported by a data center in Tampa, as well as a caching center in Amsterdam.
But the organization hopes to change all this with the build out of its second data center in Virginia, fulfilling its priority of “ensuring high site availability” for Wikipedia.
In its recent annual report Wikimedia revealed that this new second data center would “enable safe failover in the case of disaster,” as well as “increase uptime by improving site monitoring, capacity planning and operations response.”
In an email interview with the WHIR, Wikimedia CTO Danese Cooper discusses the organization’s plans for the new data center in Virginia and how it will drastically improve service for Wikipedia and its other sites.
WHIR: When is the second US data center projected to begin and when is it expected to open?
Danese Cooper: The team started researching where to locate a new data center almost two years ago, but there just wasn’t enough money in the budget to do it. Luckily last winter’s fundraiser went really well, so we’ve already started visiting possible colocation facilities and shortly we’ll be opening an RFP. We hope to bring the new data center online by January 1st, 2011.
WHIR: What makes Virginia an ideal location to build the second US data center?
DC: Wikimedia Foundation projects are important to hundreds of millions of people now, and they deserve to be hosted in the best possible way. Most of the top 10 websites in the world host at least part of their traffic out of Ashburn, Virginia because of the Equinix facility and their superior interconnectivity options. When we were visiting around in June we saw cages populated by very well known and high-traffic entities, which made us feel right at home.
WHIR: As I understand, the foundation received a $2 million grant from Google earlier this year for data center expansions. Seeing as how the new data center is estimated to cost $3.47 million, where will the additional funds come from?
DC: When you build out something big like a second data center, you figure out what it’s going to cost and you find the budget to do it. And then if you’re a non-profit like we are you work as hard as you can to find in-kind donations to bring the project in under budget (so you can re-allocate part or all of the earmarked budget on “other things”). So the entire budget for this project is already allocated by the board as part of the general budgeting process for the Foundation’s work this year, and although we allocated 100 percent of what we think it’s going to cost to build, we’re hoping to bring it in far under that amount. By the way, the generous grant from Google Labs was actually unrestricted, meaning it isn’t earmarked for a specific project. It’s gone into the general operating budget.
WHIR: How will having a second US data center in place prevent future outages?
DC: Both the outages that have happened since I joined in February were directly related to cooling or electrical failures. The facilities we are visiting in Virginia all have super efficient heat-exchange designs and very elaborate auxiliary power setups (including massive generators that kept customer sites up during complete city grid power outages as a result of “Snowpocalypse” last winter). These facilities are purpose-built for high-traffic hosting, with very robust electrical and building designs. Our mandate is to double the number of people we serve within the next five years, and to do that we’re going to need an even more stable base than we rely upon today.
WHIR: Are there any plans (or is there even need) for a second European data center after the second US data center?
DC: We don’t actually have a full data center now in Europe. We do run a caching center in Amsterdam, for which the hosting is generously donated by a company called EvoSwitch. The plan is that once we’ve brought the Ashburn data center online, we will always maintain redundancy (eg. two full data centers) so if one fails we can still serve our traffic. In the coming years we may change the configuration a bit. For instance we may add additional caching centers to serve areas where latency is still a big problem, or we may look at locating our second data center somewhere other than Tampa. We may eventually need to add another data center somewhere in Asia if our plans to stimulate increased participation there succeed. Right now it’s just really critical to upgrade our baseline.
