“Do more with less” I lived by this rule for almost 9 years now and let me tell you, I became very skillful at saving money and providing a cost effective solutions for companies. Being IT Systems Engineer I have experienced drastic changes in the world of computing in the past ten years. Just like More’s law states “The number or transistors that can be placed on an integrated circuit doubles every two years.” We achieved and continuing to improve on faster speeds in our hardware, but what about storage? How is storage planned to grow? We seldom hear of Kryder’s law and his predictions are as important and precise as More’s. Kryder’s law describes hard disk storage cost per unit of information. It predicts that the disk storage will grow in a similar rate as “transistor per integrated circuit” scenario. In a short time of a decade disk storage speeds and capacities have increased substantially. We, as a society, have accomplished and improved on many factors such as error correcting, security and reliability as far as data management goes. Let’s face it, we trust our storage media with vital information and we demand performance and reliability of it.
Well this is all great and I am sure no one wants to hear how well we are doing and what we have accomplished for the past decade. Most of us want to know and hear of what is going to happen in the future or near future for that matter. With that question in mind, I would like to share some of my own discoveries and points of view, trough the eyes of an IT Systems Engineer.
First place I have looked was the “Storage effect”. My first question was what is the Storage effect? Everyone is talking about it but no one knows for sure what that even means. I have asked a few co-workers and everyone had a different understanding of what “Storage effect” really is. Most of them referred to More’s law and some were very much confused of what others had in mind.
With those responses in tow, I went to the biggest software manufacturer in the world, “Microsoft”. I often attend TechNet events in my area as I like to stay on top of all the security and data management options which are available out there for professionals like me. The event that I had attended was structured for IT Pros and as I found out later, the event was all about “the Cloud”. Microsoft's pitch in this event was that, why bother running servers and manage storage and backup when you can pay a small fee and let “us” manage everything for you. In the middle of the event I saw some questions from the audience like “Should IT Pros be worried about their jobs? If no then what should we expect in the next 3 to 5 years?” Great questions I thought and the Microsoft team did a great job explaining that the IT staff will still be needed to operate and perform maintenance duties in the "Cloud" as well as outside of the "Cloud". (For now).
What I learned from this event was that storage and lack of it in the past projects/years is slowly phasing out of the picture from the average consumer. Just like Microsoft said: “When you have a small IT staff, the resources are limited and the price is high. With our "Cloud" computing you are gaining unlimited computing and storage power with an IT staff of thousands for a fraction of the cost.” This makes sense if you stand back and look at the overall picture of IT market and operations.
After the event I spoke with a "Cloud" computing specialist from Microsoft and we went over a few scenarios on how and what we should all expect in the next 3 to 5 years. His vision on storage was that we are going to continue to expand and populate large storage farms with storage solutions. Playing stupid, I asked if he thinks we are going to be still using hard drive and solid state drives. We both had a chuckle and came to a conclusion that we, as a society, are going to be transitioning from hard disk drives to solid state drives in the next 5 years.
I have addressed Seagate with a series of questions and they were more than happy to answer them for me. Here are some questions I asked:
What is Seagate’s interpretation of "Storage Effect"?
The article on Constellation ES was direct and to the point. When looking for a high capacity enterprise class drive, you can’t beat the Constellation ES SAS drive. SAS also provides a higher level of data integrity over SATA – in fact, 10X higher system level protection against silent data corruption -- with full, robust error detection and correction both within the drive and at the system level with the latest Protection Information (PI) standard. As you concluded, the self-encrypting drive technology takes storage to a new level providing information security where the data lives – on the drive -- and provides an added benefit for IT managers with its’ quick, easy and secure drive disposal capability.
What are the predictions (in years) of transitioning from hard drives to solid state drives for enterprise environment?
Solid state drives and hard drives are complementary. Both will exist for the extended near term with SSD's services the highest performance (Tier 0) needs of the enterprise and HDD's servicing the highest capacity (Tier 2) and balanced high performance and high capacity needs (Tier 1).
Are we near the peak of our limit for storage capacities?
We don’t see storage capacities peaking anytime soon. With each new generation of Constellation high capacity drives we trend around the need to double the capacities on each drive. For instance, last year we were shipping our 2.5 inch Constellation drive with 500GB. It’s now shipping with 1TB and the Constellation ES has gone from 2TB to 3TB.
What should we expect from Cloud based communities that offer storage services?
Cloud providers are using all of our enterprise-class drives today. Like traditional data centers, they have different application needs that require different drive performance. For instance, GMAIL and Facebook fixed content applications require high capacity Constellation ES drives while applications like Microsoft Exchange require higher performing Savvio 10K drives and our Pulsar SSD's and Savvio 15K hard drives perform well in critically important caching environments. One size doesn't fit all.
Will storage get cheaper per unit of information?
The pricing for hard drives has continued to drop as we’ve added more capacity as illustrated in the $/GB pricing you’ll find at the on-line retailers or distribution sites.
With higher capacities are we jeopardizing efficiency and or performance?
Our performance has improved with each generation (along with increased capacity). For example, the 2TB Constellation ES drive with 150MB/s sustained transfer rate and the latest 3TB Constellation ES drive offers 155MB/s sustained transfer rate. And we are always looking for ways to improve efficiency. With new Seagate features like PowerChoice® where you can reduce your power & cooling costs by up to 70% or by employing our new RAID Rebuild functionality that allows you to perform RAID rebuilds faster saving you hours if not days in lost productivity.
Seagate was the first in the industry to set the bar at a 0.44% annualized failure rate (AFRs) for enterprise drives. For our newest generation of Tier 0 SSDs and Tier 1 small form factor HDDs this translates to a staggering 2.0 million hours between failures. Couple this with the data integrity and reliability advantages of the SAS interface and new features such as Protection Information, and you can understand why Fortune 1000 organizations across the globe rely on Seagate drives to store their most critical data.
From an efficiency and performance perspective, each consecutive generation of the Seagate Savvio 15K and Savvio 10K 2.5-inch small form factor HDDs and the Seagate 2.5 inch Pulsar XT.2 and Pulsar.2 SSD drives has improved storage efficiency enabling enterprises to do more with less. Be it complete more storage IOPS within any given timeframe or store more data within any given amount of datacenter real estate. Plus – our Unified Storage architecture enables organizations to simplify already complex IT environments by standardizing on a best of breed 2.5-inch small form factor, a highly reliable SAS interface, and a common, government grade encryption technology to protect data at rest and data stored on devices needing to be retired or repurposed.
Seagate has been very generous with their answers and even provided their newest Constellation ES.2 hard drives for an evaluation. Constellation ES.2 drives are filled with latest and greatest technologies which provide a whooping 3TB of storage. These drives are strongly oriented towards enterprise level systems so you might not find these in the desktop section of your favorite hardware stores. Seagate does offer a 3TB drives to desktop level consumers in the Barracuda XT line.
Very recently Seagate broke the Areal Density barrier with world’s first hard drive, Barracuda. Seagate managed to store 625 GB on square inch of platter. You can read more on this in this press release (http://www.seagate.com/ww/v/index.jsp?locale=en-US&name=unveils-1-terabyte-platter-seagate-pr&vgnextoid=6fbdb5ebf32bf210VgnVCM1000001a48090aRCRD). The new Barracuda comes with 3 platters, 1TB each and SATA 6Gb/s interface. But what about an enterprise level system? This is where Constellation ES.2 comes in to play.
Since we pretty much moved away from SCSI interface, SAS interface seems to be the right way to go into the enterprise. Constellation ES.2 comes in two flavors, SAS and SATA. Both of these interfaces deliver up to 6Gb/s performance which is intended for fast accessing systems. Constellation ES.2 has been optimized as a near line drive for providing 24x7 enterprise reliability as well as enhanced error correction.
For additional information I went to the Seagate website. Here is where I found out that these drives are best used in the following applications:
The last application interested me the most. Cloud storage. This is what the Microsoft guys were talking about. Microsoft’s vision was that everyone will be utilizing the “Cloud” sooner or later and now that we have a technology that can take us there, it’s all just a matter of time.
Let’s take a look at Constellation ES.2 SAS and SATA more closely.
Form factor has not changed much over the years. We continue to see the same 3.5 inch form factor of hard drives but yet we pack much more in to them than we did before. Dimensions might of been the same but interfaces change rather quickly. Constellation ES.2 series drives come in SATA and SAS configurations. There is nothing different about these two hard drives as they both offer 3TB capacity with areal density of 444Gb per square inch. What is fascinating here is the MTBF or mean time before failure of Constellation ES.2. We are looking at 1.2 million hours before failure with an annual failure rate of just 0.73%. Wow! That's about 54 years!
I should probably correct myself now as Constellation ES.2 offers an additional option on Constellation ES.2 encryption. (ST33000651NS and ST33000651SS respectfully). Many financial institutions require that companies who need special clearance to be FIPS 140-2 certified. For those who don't know what FIPS is, it stands for Federal Information Processing Standard. The level 2 of the FIPS 140 states (Security Level 2 improves upon the physical security mechanisms of a Security Level 1 cryptographic module by requiring features that show evidence of tampering, including tamper-evident coatings or seals that must be broken to attain physical access to the plaintext cryptographic keys and critical security parameters (CSPs) within the module, or pick-resistant locks on covers or doors to protect against unauthorized physical access.). This standard is widely used by the US government as you would need to be compliant with 140-2 for most of the applications.
As I had mentioned earlier Constellation ES.2 SATA and SAS version of drives were provided to us for an evaluation so the best way to evaluate these hard drives is to really beat on them for an extensive period of time with enterprise level hardware. I have ended up using one of our smaller servers, HP ProLiant DL365 to run some benchmarks. The DL365 is used to run Hyper V environment and with an additional SAS/SATA RAID card we were really ready for some testing. The RAID card of choice was LSI MegaRAID 9240-4i 6gb/s. 9240 is an entry level card and should give us a good base understanding of what we should expect from these drives. The tests and benchmarks were done with hard drives with similar interfaces.
What I had to do is set RAID levels on 1, 2 or 3 drives in the array of Raid 0,1 and 5. Each time the RAID was created I would run Full Initialization of the array and then perform my benchmarks. Each Initialization takes about 5 hours so as you can imagine this took a while. The choice of benchmark software I chose was HD Tune Pro 4.01. HD Tune never pointed us in the wrong direction.
First on the list was Constellation ES.2 SATA hard drive. In RAID 0 (1 drive), the maximum read speed achieved was 157.8 MB/s, with a minimum of 70.4 MB/s and an average of 121.6 MB/s. The access time was 12.9 ms and the burst rate of 203.3 MB/s. Not too bad for a 7200 rpm hard drive. The second test was in RAID 0. This time I concentrated on write speeds of Constellation ES.2 SATA. The maximum write speed was 156.4 MB/s, with a minimum of 68.8 MB/s and an average of 119.2 MB/s. Access time was 12.7 ms and the burst rate was 219 MB/s. As you can see, the drive stand-alone performed very well and exactly to the parameters that Seagate promised. Following the Constellation ES.2 SATA drive, I have looked at Constellation ES.2 SAS in RAID 0 as well. Read numbers weren't totally off. Maximum read speed clocked was at 161.0 MB/s, and a minimum of 75.8 MB/s, with an average of 161.0 MB/s. Access time was still high at 12.6 ms and the burst rate was 289.5 MB/s. Major differences here were the burst rates. This is exactly what I would expect from SAS 6gb/s interface. Write benchmark was very close too. Constellation ES.2 SAS has achieved a maximum write speed of 157.6 MB/s, with a minimum of 76.4 MB/s and anaverage of 122.1 MB/s. Access time was 12.6 ms and a burst rate of 283.2 MB/s was recorded.
The second benchmark set was started on Constellation ES.2 SATA, but this time in RAID 1 (2 drives). This is the simplest RAID mode that provides redundancy of data, so the speeds should be pretty much the same as in RAID 0, which only has one drive. When benchmark ran, I got very similar numbers, however, access time improved. In RAID 1, SATA drives had access time of 10.8 ms and so did SAS. In write benchmark, we experienced close to identical read and write results and access times.
The third and final set of benchmarks were done in RAID 5 (3 drives). Here is where the wings opened up and we were ready to fly. Like previous benchmarks I started with Constellation ES.2 SATA. The maximum read speed was recorded at 298.1 MB/s, with a minimum of 136 MB/s and an average of 230.9 MB/s. Access time had gone up to 13.3 ms and the burst speed went down to 218.0 MB/s. Interesting behavior. This could be caused by a LSI controller but never the less the numbers are rather impressive. Write speeds of Constellation ES.2 SATA experienced a maximum 301.2 MB/s, with a minimum write speed of 139.6 MB/s and an average speed of 226.0 MB/s. Access time was recorded at 13.0 ms and the burst rate was recorded at 218.0 MB/s. Not too shabby you if you ask me. Over all we experienced a dramatic increase of transfer rate in Constellation ES.2 SATA.
Constellation ES.2 SAS in RAID 5 was next. Read benchmark performed very well. The maximum read transfer rate was 304.1 MB/s, with a minimum transfer rate being 139.3 MB/s and an average transfer speed was recorded at 228.2 MB/s. Access time was still high at 12.8 ms, however, burst rate improved to 285.0 MB/s. Impressive! Write benchmark was not too far off from read. Maximum write transfer speed was 302.7 MB/s, with a minimum transfer rate being 131.1 MB/s with an average transfer rate of 223.8 MB/s. There was no change in access time or burst rate.
Overall performance of a single Constellation ES.2 has been pretty good as it sustained speeds at about 150 MB/s. Keeping RAID technologies in mind, the more disks in array the better performance we should expect out of the array. We have proved this by setting up RAID 5 configurations with Constellation ES.2 SATA and SAS. If you an IT geek you should expect great performance out of these drives and it won't brake your budget. "Do more with less".
It is very hard to conclude on an article such as this. I have made my way in and out of meetings and interviews as I had to perform extensive research on the matter of More and Kryder's laws just to understand where are we going with today's technologies and how fast are we getting "there". With Seagate's help I got to take a look at the latest and greatest hard drives that are available in the market, which offer performance/storage/protection and reliability for pretty much any application out there. It still blows my mind in terms of how far we have advanced in the past decade, not only in the technological state but also in social states. In my opinion, the demand for storage always will be there and it will get cheaper per unit of information. Just like Seagate said: "The pricing for hard drives has continued to drop as we have added more capacity as illustrated in the $/GB pricing you’ll find at the on-line retailers or distribution sites." But do we really need all this storage? Well, there are two sides of the coin. One side is directed towards social/casual user and the other towards enterprise level applications. So in short, the answer is yes, we do need all these capacities. A simple rule of thumb is that I would rather have a lot of storage capabilities and not need it all, rather than needing the storage but not having the ability to store more. From Microsoft's point of view, we are all going to be using "Cloud" one way or the other in the near future so storage and speed of data transfer is very much crucial for the Cloud development.
I would like to leave you with the same messagequote I professed in the beginning of this article. "Do more with less".