A peek inside the Gap’s (server) closet.


The next time you buy a sweatshirt, flip flops, yoga gear or even a bridesmaid dress, you could be doing it thanks to OpenStack.

Gap Inc., the largest specialty retailer in the United States — the holding company houses brands Gap, Banana Republic, Old Navy, Athleta, Intermix and Weddington Way — runs on OpenStack. Elijah Elliott, cloud domain architect for the Gap, spoke at the Boston Summit about how the Gap uses OpenStack.


Elliott describes himself as “jack of all trades,” past jobs include data architect, database administrator, network engineer and UNIX technical support person.

“From my perspective, it was a wonderful tool for doing infrastructure delivery. Instead of having to buy a server, set up bare metal or get into VMware, find an image, update an image etc. it allowed me to do everything in one place. It literally changed the paradigm. We heard from customers that it could take five, six or seven weeks to get an app- just to get the infrastructure for their application.”

Now, servers spin up in 90 seconds, the networking on those servers is up in 90 seconds and people are working almost on day one, he says.

Currently they have five private clouds. Two clouds run on Liberty plus three legacy clouds still on Havana with workloads being migrated to the newer clouds. Elliott calls it a “multi-data center.” In addition to a compute cloud,  they have object storage, database-as-a-service and some network function virtualization. “It’s becoming much more of a consumable cloud from a developer’s perspective.”

“We have literally spun up and destroyed millions of VMs over the past four or five years,” Elliott says, adding that the number is close to 3 million now.

Some 90 percent of Gap’s forward-facing applications run on OpenStack. So if you walk into a store today, Elliott says, and they don’t have the item you want, the salesperson can pull out an iPad, complete the transaction  and ship it to your house. All of that happens — start to finish — on OpenStack.

For other those who want to get OpenStack into their companies, Elliott suggests starting with it as an infrastructure tool. “Internal infrastructure isn’t going to go away, especially if your company has brick and mortar stores, if you’re supporting cash registers, if you’re supporting your sales organization or HR organization. If you want to speed up infrastructure delivery, you can do that day one with OpenStack,” he says.

Challenges and benefits

No matter what you’re coding, no matter what tools you’re using, you can use an API, Elliott says. You can have you pipelines and your cattle in that API. The challenges and benefits – cattle vs. pets, networking, adoption and DevOps — are exactly the same. “If your challenges and your wins aren’t the same thing, you’re probably not trying hard enough.”

Cattle vs. Pets

Take cattle vs. pets, for example, the only way to fix that is to educate users. “I don’t want to tell my users how to consume this. I just want to give it to them and help them consume it.” When they have a special pet that requires a lot of care and handling, they should know that we have Chef to make it repeatable and Heat to orchestrate servers, etc. “It’s a challenge to change the way your developers or users think, but when you get there, it makes everything great.”


Networking is also a challenge for a legacy company like the Gap. His advice for people facing similar situations is to sit down with the siloed infrastructure teams and start talking about how to get to network function virtualization and software-defined networking. Elliott says that right now they have flat network today inside OpenStack with the capability to do VXLAN and SDN and are “working very hard” on getting load-balancer-as-a-service together. “The reality is our network is still a legacy network, but I wouldn’t let that discourage you. It’s a challenge.”


It was a quick win with infrastructure, he says, and definitely one of the big benefits. The challenge is a bit more subtle: “As soon as [developers] are aware that it’s out there and it’s easy to use, they’re going to use it. They’re going to use it until you go crazy because they’re over-using it.”


“DevOps requires some people with infrastructure knowledge to be in the mix. It requires them to make reasonable rules that don’t block a user from using your platform the way they want to, but makes them use it in a sane way,” he says.

You can catch his entire 30-minute presentation below.