Globus services will be unavailable for up to 30 minutes beginning at 10:00 AM US Central on Saturday, September 18, 2021. Read more about the planned maintenance.
October 19, 2017 | Steve Tuecke

This is the second in our Storage Innovations blog series, and the first post about on-premise solutions. In this article we’ve invited Matt Starr at Spectra Logic to provide insights about what to expect from turnkey solutions for on-prem storage (by “turnkey” I mean essentially plug and play devices that just work, and require very little effort to manage).

A Conversation with Spectra Logic CTO Matt Starr

[Steve Tuecke] First off, Matt, give our readers the facts they need to know about Spectra.

[Matt Starr] Spectra Logic provides affordable, tiered storage solutions that allows large data users, such as High Performance Computing (HPC) and research organizations, to affordably store data on disk, tape and cloud. Started as a tape company, Spectra has been in the data storage industry for nearly 40 years, and as storage has evolved, so has the company’s product line, to include disk and cloud offerings.

[Steve Tuecke] Thanks Matt. So -- I’m sure you saw the recent story about the tape technology from IBM and Sony that has a maximum theoretical capacity of 330 terabytes in a single palm-sized cartridge. I think this both proves that tape storage is very much alive and well, and signals a resurgence of energy and innovation around on-premise (non-cloud) storage solutions in general. What was your reaction, and what do you think about solutions like this one in the context of delivering turnkey storage for users? Do devices like this play a role?

[Matt Starr] Tape remains an ideal storage medium for many users. Its density has far outreached all other storage mediums, doubling capacity with each new generation and still remaining the most affordable option on the market. IBM and Sony’s 330 TB tape cartridge is a great example of how tape is still innovating in the storage industry.

Disk manufacturers on the other hand are having to come up with creative ways to increase capacity, from filling drives with helium to using lasers to heat the disk platter to squeeze in additional capacity (HAMR). While they are able to increase capacity, they are struggling to maintain comparable  performance.

Tape has been able to achieve superior capacity and performance growth. The IBM/Sony device demonstrates tape’s strong roadmap, ensuring that even as your data multiplies, your footprint will not.

[Steve Tuecke] Great observations, thanks. Taking a step back, let’s pick apart the notion of turnkey storage. Turnkey sometimes gets interpreted as “catch all.” Playing devil’s advocate, there are too many possible incarnations of a production storage environment for any “turnkey” solution to cover and solve everything; in fact the term itself can sometimes get rejected as not possible enough to be taken seriously. What do you think is realistic for research storage leaders to expect in terms of turnkey storage? Is the entire concept of turnkey a misnomer?

[Matt Starr] Every storage solution will have to be customized to each customer’s environment. No one solution works for everyone. Turnkey can get you most of the way there; with small modifications it can be customized to your exact needs and adapted to fit into your workflow. Take BlackPearl for instance… With Advanced Bucket Management, data policies can be set to determine where BlackPearl sends a user’s data, how many copies are created and for how long they’re stored. Taking one step further, users can utilize Spectra’s Software Development Kits (SDKs) to develop clients designed to fit the application to meet their specific needs.

[Steve Tuecke] OK, makes sense. We're happy to have BlackPearl as one of the Globus storage connectors. So with all this in mind, what are the elements needed for an on-prem solution to be truly turnkey? And what role do you see tape or archival storage playing in particular?

[Matt Starr] It’s difficult to develop a truly “turnkey” solution because everyone’s data center is different. It depends on the customer’s environment, what applications they’re using, and what they’re hoping to achieve.

An important thing for users to keep in mind is that utilizing an open-standard interface allows for a much more flexible solution. For example, if you’re using object storage interface, it’s easier to interchange the back end pieces that sit behind it, such as tape or disk. Spectra’s BlackPearl is one of few options that allows for object storage with tape on the back end. With tape as a tier-one, offload storage, users can reduce the amount of data landing on their primary storage, making it faster, more efficient and more affordable.

[Steve Tuecke] In closing, what should readers be looking for in the next 6-12 months? What should heads of Research IT / Research Computing be thinking about in their data plans?

[Matt Starr] Readers should be looking for affordable, long-term storage that enables data growth with steady budget. Data is continuing to grow, while budgets are not growing as quickly. Organizations these days need to do more with less, and pick solutions that will stand the test of time so that they aren’t having to replace storage every few years. They should be looking for cost-effective, second and third tier storage that allows them to grow in capacity without increasing overall storage spend.

They should be considering whether they’re looking for on-premise, cloud or hybrid. If they’re considering a cloud solution, it’s important that they understand the expense to retrieve their data, in order to evaluate whether or not it’s a viable option for them. On-premise solutions on the other hand provide unlimited access to data.

They should also be thinking about “share-ability.” Data is essentially a currency for a lot of research institutions. They need the ability to share, distribute and collect data from other organizations so they can analyze/run analytics on data – which is where BlackPearl and Globus shine with the ability to share data to any Globus user, while maintaining an affordable turnkey solution. For organizations creating irreplaceable data, it’s essential that they are able to produce multiple copies of their data. If they are able to re-compute, only one copy may be necessary.

Up next:

Great conversation, thanks Matt. In our next article (3rd in the series) we’ll dig into the pros and cons of “roll your own” storage for research organizations -- stay tuned and thanks for reading!