Cloud computing is experiencing strong sustained growth across the industry, and the relevance of the emerging cloud market to storage and how the cloud infrastructure is evolving is consequently driving increased demand for storage products.
These Q&As address how Seagate, the world leader in storage devices, plans to strengthen the cloud computing infrastructure and how it will manage environmental operating conditions, workloads, data protection models, security and new storage requirements.
Wes Purdue is the director of enterprise PLM cloud strategy at Seagate Technology. He recently spoke at WHD Global 2012 in Europa-Park Rust, Germany.
Q: When it comes to the cloud computing infrastructure, are there differences in the data protection model?
Wes Purdue: In traditional IT, RAID has been the predominant data protection model. RAID is also used in these cloud infrastructures, but in these large infrastructures, it’s typically not a hardware RAID that’s the predominant data protection model. It’s replication, and it’s at least 3×. As we develop storage devices over time, we try to address features for RAID rebuild to help improve that—for hot swaps, for hot swapping drives, hot plugging drives—but quite frankly, not all, but a lot of the Tier 1 cloud service providers say that they replicate, so these features, we can’t fully utilize them.
Q: What features can’t fully be utilized?
Purdue: Seagate offers full disk encryption, and we have self-encrypting drives, but the cloud guys really can’t use that. They come back and say, “as long as the drive is in our data center, we feel the data is pretty well protected. We have what we call guns, guards and gates.” Seagate has this great advanced encryption scheme, AES-256, data-at-rest, but cloud providers tell us: “We really can’t use it that way, and we definitely don’t want a key management system that is complicated, especially in this large scale-out.”
Q: But what happens when the drives leave the data center?
Purdue: That is a challenge. When a drive leaves the data center, there are two primary scenarios: 1) the drive is repurposed and goes to another site, and 2) the drive is retired and destroyed.
Typically the process is that a third party outside of the data center destroys the drive. Of course the data center will protect the data at all costs before that drives leaves the site. The traditional process has been to wipe the drive, and that takes about 13 hours for a 3TB drive. Typically they wipe it at least three times. That’s 39 hours, so you’re looking at days.
Q: Is there an alternative to the traditional process of wiping the drive?
Purdue: Yes. In fact, Seagate came up with a way to do an instant secure erase in less than a second. It has a lot of value, especially to these large Internet data center build-out deployments, and it’s consistent with their current security process. Instead of taking 39 hours to wipe a drive to secure that data, they can cryptographically erase the data in less than a second. This is another example of taking security technology—our encryption technology—and redeploying it to a feature that fits the need in the cloud space.
Q: What about the need for cold storage?
Purdue: We are hearing from a lot of the large Internet data centers that their customers are saying, “don’t delete my data, at least not now.” For business purposes and for regulatory requirements, there’s more and more data. But even in the cloud, there is archive data, cold data. Service providers are very good at migrating data, and they are asking for a way to transparently migrate data from our warm systems to a cold system, and then back up, bi-directionally, with automated migration. Here is another need in the cloud space in terms of providing a system, providing a storage device, that has the attributes to support this type of a cold storage need. Seagate can definitely help here as well.
Q: From a Tier standpoint, how are Seagate products distributed across the cloud?
Purdue: Tier 0 is the high-performance tier, Tier 1 is a balance between performance and capacity, Tier 2 is the capacity-optimized tier, and Tier 3 is archival-optimized. Across all infrastructures, these tiers are used. It’s not just a single storage device. There are a number of applications that fall into each of these four tiers.
From a product perspective, whether it’s an SSD; a Seagate® Pulsar® drive or Savvio® 15K- or 10K-RPM drive for the compute type of applications; or perhaps a Constellation® or Constellation ES 2.5-inch, 3.5-inch nearline/business-critical drive with 3TB and 4TB capacities, all of these types of storage devices are used across all different applications. Depending upon the workload, the performance, the capacity and the way the platform is architected varies and will depend on which storage device is optimal for that use.
Furthermore, take it outside of the enterprise and you have small- to medium-sized businesses with NAS devices creating their own little cloud, personal home clouds, and even mobile clouds. And those storage devices have a further set of attributes that have to be satisfied for those particular applications.
Q: Finally, what are the keys to success in supporting cloud services?
Purdue: First and foremost, dedicated support is critical. In some cases that means having an engineering team—badged—on location at the customer site. Local high-touch technical support is very critical to these service providers. However, what I am seeing is that the smaller builders and integrators understand this need more than the bigger guys.
Read Part 1 of this article