Paula Long is no novice when it comes to enterprise storage. In 2001, she co-founded EqualLogic, a storage-area network vendor that Dell bought for $1.4 billion in 2008, and now she’s back in the saddle with a new stealth-mode startup called DataGravity. She’s also not sold on the viability of the flash storage market, which might already be reeling after Violin Memory’s subpar IPO on Friday and a couple others in the works.
Long shared her opinions on flash, cloud computing, software-defined networking and more our Structure Show podcast this week. If you’re into smart commentary about what storage technology is and should be, you’ll definitely want to listen to the whole thing (download options are at the end of this post). But here are the highlights.
CIOS, spend your money wisely
Long acknowledged that today’s flash storage systems are very fast and efficient, but the problem from her perspective is that most applications simply don’t need that type of performance. Right now, VDI might require flash, as might an OLTP system doing 1 million transactions a minute.
“If you need to go zero to 60 in 2 seconds, flash is probably a great way to get there,” Long explained. “If you don’t need to go in 2 seconds, it’s probably an expensive solution for what you need. Storage has never been one-size-fits-all and it never will be.”
She also advised against buying into the pitches from certain flash storage vendors that they’re now price-competitive with disk, something that would make flash look like a good investment even if you don’t need the Ferrari-like performance. “Basically what they’re saying,” she said, “is, ‘If you have the perfect dedupe ratio and the perfect app, it gets close.’”
Flash vendors, look to the future
Assuming IT buyers don’t rush to buy flash systems, Long questions how much of the small addressable market any given flash vendor can realistically hope to capture. EMC, HP, NetApp, Cisco and IBM are all selling flash arrays alongside disk-based systems, which means flash startups could have their work cut out to carve out significant businesses.
Long said the biggest opportunity for flash might actually come two generations down the line, when applications should be able to address it as memory rather than storage. “That’s when it gets really cool,” she said, “because the applications you’ll be able to build on that will be something we’re not even imagining yet.”
Standalone cloud storage is a tough sell
Had we spoken with Long a few weeks ago, before the news that cloud-storage pioneer Nirvanix is closing shop, her thoughts on the subject might have seemed prescient. Essentially, she argued, it’s difficult to reap the rewards of standalone cloud storage because users might lose control over the things they’re trying to solve in the first place.
“If you have a child who’s sort of not behaving — you don’t know how to control it and you want to reduce spending — sending it away to boarding school so you don’t have to watch doesn’t mean it’s less expensive, or going to behave any better when it came home,” Long analogized.
If companies are going to move their storage to the cloud, she said, they’ll probably be moving their applications with it. That’s why the software-as-a-service model works so well already. “It’s not where the data lives,” Long said. “It’s where the data and the knowledge that knows how to work on the data lives.”
Goodbye, big boxes
When it comes to how storage of any kind is delivered, Long said she’s pretty confident we’ve seen the end of the big-box era — at least from startups. She thinks 3PAR, which HP acquired for $2.4 billion in 2010, was probably the last one to pull that off.
Here’s what Long had to say when I noted that Oracle is still selling really big boxes: “It would be interesting to know if they’re building very big boxes for their installed base or if they’re building very big boxes that are breaking into new markets. I can’t answer that for you, but I can suspect what the answer would be.”