There is no longer any doubt in my mind that tape will be around long after I am gone. Not only did Spectra Logic announce a new tape library (the T-Finity) that is targeted at the very largest of enterprise accounts this week, but a disk storage system representative made the tongue-in-cheek comment that partly serves as the title for this week’s blog while we were talking about the possible sunset of specific disk and tape technologies.
First things first. To say I was bit taken aback by Spectra Logic’s announcement that it was coming to market with a new enterprise tape library is an understatement as the announcement was as unexpected as it was shocking. Because after all of the dire predictions that tape is on its last legs, along comes someone who puts a stake in the ground who says that not only is tape not dying, but realized that customers still have a need for tape and recognized a market opportunity.
Spectra Logic is not going crazy about the potential for this product. There is only a small segment of the market that actually needs a tape library product of this size that supports up to 8 robots, 100 frames, 122,000+ slots or 480 tape drives with a base configuration price that starts at over $200K.
However those few organizations that do need tape libraries of that size face a serious problem. Prior to Spectra Logic’s entrance this week into the high end tape library market, there were only arguably two providers in this space – IBM and Sun StorageTek.
The price point of IBM’s tape libraries limit its appeal to those organizations that have the GDP of a small country while the future of Sun StorageTek’s line of tape products continues to remain cloudy. This lack of clarity about the future of these products, their high price points and the lack of innovation in current enterprise tape libraries gave Spectra Logic an unexpected opening to make inroads in the highest end of the tape library market.
While Spectra Logic has only one pending sale for the T-Finity that it discussed on the phone with me, its arguments for the T-Finity finding new markets for growth make sense. Unlike IBM and Sun StorageTek, Spectra Logic is looking to go after the markets where lower costs and innovation are paramount.
Specifically, it has identified the high performance computing (HPC) space as well as the entertainment and media spaces as specific verticals that need lots of storage capacity at a low price. In both of these environments, the idea of information lifecycle management (ILM) is a bit of anomaly since they essentially expect to retain all of the data they create forever.
So by Spectra Logic coming to market with a tape library that minimizes the up front costs, gives companies the flexibility to scale and makes better use of existing data center floor space than competing systems while making a commitment to support its tape library, one would think Spectra Logic’s T-Finity immediately has to go on the short list of every company (absent mainframe shops) that is seeking a tape library of this size.
Spectra Logic’s announcement also came up later this week while I was having a conversation with Dot Hill’s product marketing manager, Scott McClure. We were discussing Dot Hill’s Nov 4, 2009, announcement regarding its new option for inclusion of STEC solid state drives (SSDs) on Dot Hill’s iSCSI and FC storage arrays.
More proclamations have been made lately about the eventual demise of 10 and 15K RPM hard disk drive (HDDs) as SSDs become more popular. However announcements such as Spectra Logic’s would seem to suggest that if there is still a future in tape, a similar fate may await 10 and 15K RPM HDDs. McClure replied, “I see the same outcome for 10 and 15K RPM HDDs that I see for tape and tape has been slowly dying at a growth rate of 2-3% per year for the last decade.”
McClure went on to explain that for now he does not foresee the imminent decline of 10 and 15K RPM HDDs. Dot Hill still has plans to support 10 and 15K RPM HDDs for the foreseeable future, especially since the cost per GB for SSD is still a 10x factor when compared to SAS drives ($2-3/GB for SAS; $25/GB for SSD).
However even McClure said he has been surprised at the uptake that he has seen among Dot Hill’s customer base for SSD. One of Dot Hill’s specialties is in developing more rugged storage systems intended for military use so the fact that SSDs are less susceptible to vibration than HDDs and can operate at high altitudes looks like it will be a boon for Dot Hill in this respect.
The types of orders that Dot Hill is getting for these units also are exceeding its expectations. Since SSD maxes out the performance of the RAID controller at 6 – 10 drives, Dot Hill expected this size unit to be the average size of the storage system ordered. It is selling many more systems with 30-40 SSDs in them than it ever expected though Dot Hill is still not entirely clear what is driving this customer demand for such large SSD systems though users needing to online file downloads seem to be driving some of it.
Finally, there were a couple of other items that I did want to mention as part of this week’s recap.
First, Estorian finally announced this week that it has hired Kevin Riegelsberger as its new CEO to replace its former CEO Gary Tidd who resigned this past summer. I mentioned in a blog that appeared back in August but was asked by Estorian not to mention his resignation and the new CEO until a formal announcement had made by Estorian so now that the press release is out there, I can pass this information along.
Second, I mentioned in a blog that appeared earlier this week that I received an email response from EMC/Data Domain’s Frank Slootman in regards to his thoughts on an interview that he appeared on SearchDataBackup.com. However in that blog, I only posted some of the email response. Here is the full response that EMC/Data Domain provided back to me in regards to the accuracy of the interview and the statements he made:
“I was not accurately quoted in every instance of the write up, some of my statements were not nearly as colorful as they appeared in print. That said, most of the positions represented in the write up are substantially mine.
On the Disk Library, we are not harvesting dedupe technology from Data Domain and grafting it on to the Disk Library for the reasons mentioned. On the other hand, and the article mentions this, we are looking at methods to migrate data from a DL to a Data Domain Restorer. This may end up helpful to off load disk libraries that need to free up space and/or vault backups off site over a wide area wire.
Data Domain Restorers are effectively a new class of Disk Libraries, and they will usurp some – not all – of the range of workloads previously thought of as pure VTL domain (no pun). We think some DL customers will just keep doing what they are doing
without dedupe, so
me will welcome integration/migration options with Data Domain, and others yet will instead opt for an all Data Domain approach.“
That’s it for this week! Have a good weekend!