There has been a quiet revolution in storage happening and for once, it has not been led by the decades-old companies catering exclusively to on-prem deployments.
Oddly enough, it has been sparked by the major cloud vendors who are finally moving beyond the basics of scalability, flexibility, and cost control to capture a far greater share of storage by adding robust file services to their ranks.
For those out of the storage loop this is critical because most of the world’s mission-critical workloads (with the exception of cloud-native applications that underpin web services) are file-rooted, a fact not likely to change since it would involve rewriting time-honed applications to speak object. Yes, there are services designed to make that process easier but for established companies with legacy codes this would be a major undertaking that could introduce risk, cost, and so on.
The cloud majors have recognized this for some time but there was no momentum in catering to this high value and high-volume base of users until recently. It wasn’t until 2018 that a lot of this kicked off with AWS leading the pack on file services additions and Azure and Google Cloud Platform (with its acquisition of Elastifile) following in short order. Now, halfway through 2019, these are catching on, which means more customers who were limited to on-prem due to file-based storage demands will start looking to the cloud.
We will be inadvertently talking a lot about cloud at The Next I/O Platform event on September 24, in part for these reasons. For instance, our chat with the head of storage engineering at LinkedIn live on stage will touch on some of these new capabilities. And the morning keynote to kick off the storage part of the day with Andy Watson, storage pioneer and current CTO at WekaIO, will focus on this topic rather extensively, setting up the conversation about the shifts happening in storage at scale, especially for large enterprises with established bases of applications.
“Among the various clouds, there is a growing focus on providing a competitive, meaningful, high performance file storage alternative that in the past they had neglected in favor of scale, elasticity, and hitting ever-lower price points,” Watson explains.
“In the past, if you were doing unstructured storage, file services, which are still dominant for the most familiar use cases, there was not a credible alternative in the cloud. The options were weak and if you tried to improvise there were a lot of tradeoffs. That’s no longer completely true. There are certainly things that are better on-prem but the argument is not as strong as it used to be and it will be going away soon.”
Watson’s keynote will reflect on what 2020 will hold when it comes to cloud, hybrid, and on-prem storage and how the shift that started in 2018 will play out over the next several years.
“2018 was a year of admission on the part of the cloud companies that they were lacking in this area and a declaration that things would get better. In 2019 it all began to happen. By 2020 we will reach a maturation point where the customers who didn’t want to be the first on the block to implement something immature can finally proceed, at least on AWS, which has a fair number of miles in among the early adopters as well as those who found out what we’ve been doing all along,” Watson adds.
Moving to file services via WekaIO or any other number of third-party managed services might be new but it is not necessarily the big leap many new cloud approaches were years ago. “A lot of folks who work on cloud-native architectures drag their feet here and say that everything is going to object and everyone should just write code to talk directly to an object store. But what they are forgetting is that there is an incredibly large body of existing applications, basically anything anyone does with computers, from standard databases to most enterprise applications. All of these are file based and that wide breadth of users will need to go forward using files, or at least many will.”
Watson says that from his own conversations, much resistance to moving to a cloud/object approach was because of the overhead and risk of rewriting applications in a cloud-native way to interact with object. He adds that the efforts in that direction have been limited. Further, one reason is that for small objects, that storage method is not intrinsically fast. “You can get great performance from a file system if a file is small or large if it’s been done properly. But even with the best object storage there is always the small object weak point unless an application has sustained, sequential, streaming access to large objects. The problem is, a lot of applications don’t have those characteristics, so file is here to stay and in the cloud, not everything can or will become a cloud-native object approach.”
It might be surprising to hear WekaIO talk up the rise of the cloud for a new set of old workloads but actually, the company was blazing this trail in advance and has products for on-prem and cloud alike. Watson says that this shift is so important and dramatic that it cannot be ignored and should be talked about freely, which we appreciate since the event will be focused on what’s next in storage and networks and more important, why is it next, for whom, and in replacement of what, and with what considerations top of mind (performance, scalability, cost, usability, and so on).
Join Andy Watson and many others at The Next I/O Platform on Tuesday, September 24 as we kick off the storage part of the day first thing in the morning. We’ll be hearing real-world storage stories from companies like Netflix, LinkedIn, Fred Hutchinson Cancer Center in a conversational interview-based format with deep dives into networking topics the second half of the day.
Our thanks to WekaIO for being a Platinum Sponsor of the event.