Solution provider takeaway: Learn which green storage technologies will reduce your clients' power consumption...
and help them better allocate their data center resources.
One of the mistakes many integrators make is assuming that customers mean the same thing as you do when they mention a new technology. Even worse, you might assume that customers will ask about it if they're interested in it. Green storage is one of those topics that integrators oftentimes don't feel like they need to get out in front of, believing that customers already know everything about greening their data center and the role storage plays in the process.
The truth is they don't; the evidence of this is in the focus of most green efforts: virtualization, usually server virtualization exclusively. What does server virtualization really do that facilitates green data centers? It consolidates. Power savings most often come from the reduction in the number of physical servers, not from the sleeping or powering off of servers. Storage is different -- while there is a consolidation effort in storage through the use of storage area network (SAN) and network-attached storage (NAS) systems, the top goals of SAN and NAS are better optimization of resources and easing the storage management burden. While power savings may be a byproduct, it is not typically a specific objective.
To get your customers to their greenest possible posture, you'll need to focus on two key areas: better capacity use and being able to power down or turn off storage.
Improving capacity management
There are several techniques that provide better capacity management. At the front lines is thin provisioning, the ability to allocate storage to the maximum needed by the application without actually using that storage until data is written. For example, if your customer has a database administrator (DBA) who wants to deploy a new Oracle application, it will often request what the project will need when the application is in full production -- perhaps 500 GB -- but it will take at least a year to get to full production. In fact, for the first eight months, this application is going to need less than 20 GB. Why waste 480 GB of disk capacity? With thin provisioning, you allocate the 500 GB to the Oracle application, but it consumes actual capacity only as data is added and moves toward full production.
If you aggregate these savings across all the servers connected to your customer's SAN, the delta between actual used capacity and allocated capacity becomes quite large. Most of the time, suppliers will focus on the hard cost savings in not having to buy as much storage, but there will also be huge power and cooling savings. Every shelf of storage that your customer doesn't buy today represents savings on the power and cooling of that shelf later.
Thin provisioning was revolutionized by companies like Compellent and HP, and most storage manufacturers have jumped into the market claiming to do some form of thin provisioning. When evaluating options for your customers, make sure that the product you're looking at truly enables thin provisioning. There should be no allocation minimum, and growth of a provisioned volume should not have to be done in certain size increments. And make sure that the expansion does not require a reboot of the server assigned to that logical unit number (LUN) or any operation on that server. If the supplier requires these steps, the product doesn't offer true thin provisioning and therefore loses some of its green storage stature.
The second capacity management approach to consider is reducing data on primary disk storage via data deduplication or data compression.
Data deduplication is a process that looks for similar byte streams within files to identify similar files. Once identified, they are stored only once. Deduplication has been very popular in the backup market, but its value in primary storage is questionable for several reasons.
One of the problems is that data deduplication requires a background process investigation of the file system to identify files with duplicate byte streams; this background process can be very resource-intensive, so it can only happen when the storage is at rest. Since the data isn't active, it doesn't necessarily make sense on primary storage -- primary storage is more cost-effective when used for active data. Beyond that, the big problem is that primary storage has a much lower percentage of duplicate data than backup data. That's because with backup storage, you purposely save the same data over and over again -- weekly full backups, for example; in primary storage, on the other hand, your customer may have some duplicate databases but not much else. There are some gains to be had here, but don't expect the same data reduction ratios that you'll see in backup operations; 2.5X to 3X is typical.
Active data compression is more interesting. Companies such as Storwize offer an appliance that sits in-line and compresses active data. Compression ratios will vary, but a 2:1 ratio can be expected for most data types, and some, like Oracle databases, will compress even further. Because it is in-line, all data is compressed while active or near active. But what makes these solutions interesting is that there is no measurable performance impact from their use: They're designed to work in file server environments (CIFS or NFS). The green impact is significant -- not only is active storage utilization cut in half but so is disk backup storage, because the data is in a precompressed state, further optimizing data deduplication backup devices.
Better allocation of the disk resources themselves will also help maximize data reduction. While the actual wattage difference between a 1 TB SATA drive and a 300 GB Fibre Channel drive is marginal, the cost difference on a watt-per-capacity basis is significant. If you can assist your customer with the movement of data to these higher-capacity drives, you can help reduce overall drive, shelf count and power consumption.
There are several methods of migrating data to higher-capacity drives. Companies like Compellent offer storage solutions that are capable of moving data at a block level to different classes of storage as the data ages. This means that even subsections of databases can be migrated from high-speed, low-capacity Fibre Channel drives to lower-speed but higher-capacity SATA drives. This makes for a simple and effective solution to maximize watts per capacity.
You should plan to eventually move data to an external disk archive designed for long-term retention and power management. While a variety of vendors offer software to handle moving the data, up until now it's been complicated and expensive. The good news here is that the software is getting simpler and more cost-effective -- companies like Atempo and Enigma Data Solutions offer straightforward data movement solutions designed for disk-based archives.
Controlling power usage
Once you've given your customer the ability to move data in a green manner, consider a disk archive solution that offers power-managed storage. The most significant power savings you can provide are those derived from powered off but ready-to-use storage. Consider power-managed archive solutions like those from Copan Systems or Nexsan. Make sure you look for solutions that offer densely packed capacity, can scale (500 TB-plus archives will not be uncommon in a few years), and can power off drives rather than just spinning drives down (nothing is more green than off). Also make sure that the system doesn't require you to provision full power and cooling to the unit. Some systems must be equipped to handle a complete power-up.
When you consider that most studies indicate 80% of data has not been accessed in the last 90 days, moving that 80% to a power-managed archive can create a significant reduction in power usage. The remaining 20% can be reduced by the use of thin provisioning on block storage and using in-line compression of file storage, making it possible to cut storage power and cooling use by as much as 90%. Not only will that green storage approach get your customer's attention, it will make you a hero.
About the author
George Crump is president and founder of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. With 25 years of experience designing storage solutions for data centers across the United States, he has seen the birth of such technologies as RAID, NAS and SAN. Prior to founding Storage Switzerland, George was chief technology officer at one of the nation's largest storage integrators, where he was in charge of technology testing, integration and product selection. Find Storage Switzerland's disclosure statement here.