Why would anyone intentionally move to a 10K rpm speed drives over 15K rpm? Especially for mission critical apps? A little analytics gives us the answer.

The key differentiators boil down to density and price. These two attributes work together to overcome the IOPS delta between 10K rpm vs. 15K rpm and here’s the secret sauce:

1) Density - 24 x 2.5” 10K rpm SAS drives fit into a 2U disk shelf, which makes it denser than regular 3U disk and even the so-called “hi-density” 4U shelf which HDS also offers.

2) Price – The individual drive and shelf price for 2.5” 10K are the same as that of the regular 3.5” 15K drive and shelf, and about a 3rd the cost of one hi-density shelf. But since 24 can fit into one 2U tray instead of 15 in a 3U tray (or even 38 in a 4U tray) the overall cost per tray is less. With the lowest cost available for SAS disk from Hitachi Data Systems (HDS), clients can easily afford to add a few more disks to make up the IOPS difference and still save money.

The following chart illustrates how this works (click to enlarge):

• Each row in the above chart represents one full tray’s worth of disk for each disk type. The top section compares IOPS per tray, IOPS per rack unit, etc… while the bottom section compares environmentals
• The “IOPS/tray” column shows numbers that favor the 3.5” hi-density 600GB option. But since the 2.5” option is 2U instead of 4U, the “Computed IOPS/Rack Unit” column levels the playing field. And since the cost is less per tray, the “IOPS/dollar” is also about even. The “Raw GB/$” is quite a bit better for the upstart 2.5” 10K drive.
• The environmental comparison is off-the-charts in favor of the 2.5” 10K. For those that have power cooling cost concerns, this is the option to go with.

What does this mean in plain terms?
We are working on a solution currently where the client asked for an AMS2500 with about 87TB usable for a hi-performance production environment. We configured a 600GB 3.5” 15K “hi-density” version as well as a 600GB 2.5” 10K version. The 2.5” 10K MSRP hardware total came to $480K while the 3.5” “hi-density” version MSRP total came to $629K, 23%(!) higher.

The total IOPS for the 2.5” 10K version is 8%-10% less than the “hi-density” version, so the IOPS delta is negligible. And since the cost is so much less, we could add more drives into the data storage solution to make up the IOPS difference while still remaining at or below the cost of the “hi-density” solution (which would also increase the usable capacity).

It’s also interesting to note the 2.5” 10K version takes up a total of 22 rack units while the so-called “hi-density” version takes up 28 rack units. Lastly, the power/cooling numbers are 46%+ less (!) in the 2.5” 10K version. Remember this is just an 87TB usable solution and the argument is compelling. It becomes even more compelling as the total capacity goes up.

This is why several of ASG’s largest HDS clients are making the switch now to 2.5” 10K rpm drives as a standard for production or are seriously considering it.

Steve Wine, our most senior account executive based out of our Denver, Colorado office summed it up:

“We presented these drive performance numbers to my largest storage client in Colorado and it convinced them to make the switch. They deliver cloud storage and other related services so the need for high-performance was paramount and we took that into consideration. We now have several AMS units installed (at the client) with all 2.5” drives and they are performing perfectly, even in their highest performance tier.  Because they buy so much disk storage, switching to 2.5” drives has been a huge cost savings both on cap ex and op ex side. This will be their standard for most modular storage going forward.”

The complete analysis has garnered a lot of great feedback from our clients who really use it for data center planning. For a copy of this analysis containing a complete Hitachi modular drive comparison chart containing all available options, contact Anthony Sayre at “asayre (at) virtual.com”.

About the Author
Anthony Sayre
ASG Advanced Product Specialist
asayre (at) virtual.com

DISC produces a comprehensive line of automated Blu-ray optical archival storage library products with online capacities ranging from 1.5TB to 70TB and more. The optical Blu-ray technology was appealing to us because of the benefits that it affords, including:

• Longevity
• Readability
• Authenticity
• Operation costs
• Environmentally friendly

We’re confident that they can deliver on the need for long-term retention of digital data while including a 50 year media life, scalable design, and true green storage.

The addition of BDXL Blu-ray data storage automation technology to our product line helps satisfy the growing demand that we see for automated, cost-effective tiered data storage solutions for companies of all sizes. It will allow us to provide customers with the scalability to drive the overall value proposition and ROI from data inception to archive.

Howard Marks raised an interesting question in his recent  blog post “So, What Is A Backup?” in Network Computing.  Backups as a Storage Tier?

We’ve been working on this concept with a lot of our clients. More than 95 percent of data loss happens at the local site. Fifty-six percent are soft errors, including patching to virus to hackers.* Why not just move the snapshot files that are candidates for being backed up to a separate tier of storage? You can keep these files in their native format and organize them in time-coherent views.

Users can restore files themselves from any point in time using a search engine. There’s no need for backup software, and it’s simpler to deduplicate and compress. You can also apply compliance and regulatory rules to it as policies, or use it as an archive. Data mining, replication, or other data moving requirements are more simply met, and indexing is easier.

Copy-based backup is now a default paradigm for consumers. Originally this method was cost-prohibitive, but not anymore. Solutions now cost the same as LTO4 media! Hence, disk is the new tape…

Even if you didn’t know anything about your data, would you know the best way to classify it?

Your best bet? Build an archive: This will quickly identify your “working set of data.” By using a file virtualization solution across the storage infrastructure, we’ve seen up to 50% cost reductions in disk and maintenance of a backup infrastructure AND a reduction in backup windows. (In one case we saw it decrease from 14 hours to 3 hours.) That means disaster recovery is vastly simpler and a great deal faster!

We’ve deployed other solutions, like Synthetic Fulls, to accomplish similar results. Backups have greatly reduced I/O loads on clients and improved the full server recovery process. Synthetic Fulls creates new backup sets offline without touching the source data or communicating with the original host server. This improves the ability to create new full backup tapes for other uses such as vaulting or setting up a new site or test system. Replicate and replay.

We’ve been working on this concept with a lot of our clients lately. Why not just move files that are candidates for being backed up to a separate tier of storage, keeping them as files in their native format, and organizing them in time coherent views? Originally it was because of cost, but not anymore. Solutions now cost the same as LTO4 media!

Users can restore files themselves from any point in time using a search engine. You don’t need backup software to do this, which means it’s simpler to deduplicate and compress, apply compliance and regulatory rules to it as policies, and use it as an archive. Indexing is easier, plus data mining and replication or other data moving requirements are more simply met. Copy-based backup is now a default paradigm for consumers, i.e. Apple’s Time Machine and EMC’s Mozy.

If you’re looking into this or other data protection or backup solutions, check out the Active Archive resources from our partner, Spectra Logic.

Initially, data deduplication eliminated data redundancy in specific cases like full backups, email attachments, and VMware images. However, you’d soon notice the pervasiveness of duplicated data. That’s because test and development data multiplies across an organization over time. Replication, backup, and archiving create multiple data copies scattered across the enterprise, and users often copy data to multiple locations for their own convenience.

Organizations now recognize that—far from being a niche technology— deduplication should be an integrated and mandatory element in their overall IT strategies.

There are essentially two ways to reduce the cost of your data storage. First, you can try to leverage a lower-cost storage platform, which results in an additional set of problems. Your other option is to leverage data deduplication to reduce your data growth and total required storage.

Data deduplication can lower the cost of your data storage by reducing the amount of disk needed to store your data, whether it’s backups or online primary production volumes. In this three part blog series, we’ll discuss five best practices to help you select and implement the optimum deduplication solution for your environment.

Like disk-to-disk backup or server virtualization, you don’t want to evaluate deduplication as an isolated product or feature. You must consider the broader implications of deduplication within the context of your entire data management and storage strategy.

For example, deduplication can be performed at the _le, block, and byte levels. You’ll have to consider the tradeoffs for each method, which include computational time, accuracy, level of duplication detected, index size, and in some cases, the scalability of the solution.

Also, consider how you can use deduplication to eliminate tape where it makes sense in your environment. That might be remote offices or any locations where your company doesn’t have trained IT personnel.

We’ll discuss two more data deduplication best practices in our next blog post, so come back soon. And please share your own best practices with us as well.

Data retention policies are a critical component of data archiving. Once you have a data retention policy, enforce it for all the information you have on your network. You should also record the retention periods, both for distribution to users and as part of the legal defensibility record. The retention schedule does not need to be elaborate, but it should include specifics about the various categories and associated retention periods. When creating a written schedule, organizations should focus on two main criteria: the maximum retention periods and data distribution.

As part of a data retention policy, you should set minimum retention periods and maximum retention periods, in order to avoid over-retention. You can either include specific maximum retention periods for each category, or you can include a general clause that when the minimum retention period expires, the company no longer retains the data. This maximum retention period should also address the disposal of any currently retained data, including information archived under an interim infinite retention period.

Once you establish your archival retention periods, distribute the information to all users, both to communicate policy and to address the storage habits of individual users. In cases where individual employees use a variety of different email programs or in cases where it is common for users to archive e-mail messages, the retention schedule should make clear that this is not “retention” for purposes of the organization.

Organizations can meet legal preservation obligations with interim archiving, regardless of whether a complete retention policy is in place. Litigation requires that a company be immediately able to preserve and retain information (also known as instituting a legal hold) as soon as it reasonably anticipates litigation or governmental action.

Simply moving information into an archive does not constitute effective information management. In order to gain the highest benefit from archiving, companies should categorize all archived information, setting retention periods for each category. Well established retention periods with automated, effective, and easy-to-use archiving systems create streamlined, customized, and uncomplicated data preservation systems.

We hope you’ve enjoyed our series on data archiving and have found it informative. Let’s recap below the 5 data archiving best practices we’ve discussed over the last 5 posts.

Data protection and privacy continue to be a tremendous focus and risk for IT communities. While companies make great strides protecting data privacy in production application environments, they often overlook implementing similar strategies in non-production environments such as testing, development, and training.

Reliable, safe, and effective mechanisms for securing data at rest require the adoption and rigorous execution of well-defined processes for handling keys used to encrypt data and keys used to safeguard the data encryption keys. Key management is a comprehensive term that covers these controls—including the creation, distribution, deployment storage, transmission, and destruction of keys used to encipher data.

We’d love to hear what your doing to ensure the security and privacy of your data, in production or non-production environments. Leave a comment or send me an email!

In our next, and last, blog in this 5 part series on data archiving we’ll discuss establishing a data retention policy.

When you implement information lifecycle management (ILM), you effectively store your information in a manner consistent with the value of the data. All data maintained on storage networks has a defined lifecycle. This lifecycle identifies the way information travels through an organization from its creation to its archival and removal. The exact steps in data lifecycles largely depend on organizational policy, though data generally travels through three stages:

1. Stage 1: Creation/Acquisition of Data. During the creation of data, both data availability and data value are extremely high.
2. Stage 2: Publication. The value and availability requirements of published data, whether printed or accessed through other means, often depend on the content of that data.
3. Stage 3: Retention and Data Disposal. The length of time an organization archives and retains information depends on the nature of the data. However, increasing federal regulations, standards, and compliance measures often govern how long organizations must keep certain types of data.

The changing importance of data and the requirement for data availability create problems—it is costly to store all information on expensive, high-availability Tier 1storage systems. At some point, organizations must shift at least portions of corporate data to less expensive storage media. However, that merely raises the questions, “What data do we move to cheaper media? And when do we move it?”

ILM provides a strategy for data management throughout the information lifecycle. It identifies the processes and technologies that determine how data flows through an environment. Information path management is another consideration with ILM organizations are unlikely to offload all rarely used data to cheaper storage unless they can still access the data reliably, if needed.

With an effective email archiving strategy, companies can often save as much as 50% on backup costs. The average employee sends and receives more than 140 emails each day, according to a recent Osterman Research report, and according to courts and regulators, emails are official business documents. Motivated in part by the Federal Rules of Civil Procedures, as well as case law emerging around email discovery, companies now have new and increased responsibilities to control email in response to potential litigation.

In order to address this issue, organizations often impose strict mailbox quotas, limiting the amount of email any single user can save. Unfortunately, limiting email storage only encourages users to store email messages on their own desktop or laptop in PST files. This strategy presents complications if companies need to provide litigation holds on this data. Additionally, storing email messages on the corporate server negatively affects its performance over time, so it takes longer to recover in the event of a disaster, and is more likely to need additional capacity and other upgrades to accommodate the increased load.

Be sure to stay tuned for the next installment in this 5 part series on Data Archiving. Our next post will discuss Information Lifecycle Management.

Archiving is largely about control—knowing precisely what data you have, where you have it and having it at the right place, at the right time. Over the next few weeks we’ll explore 5 best practices and tips to trim down your essentials and archive your rarely used content.

When you’re designing and implementing a data archiving solution, a good first step is to establish a data management baseline of your current environment. By identifying and classifying the various data pools within your organization, you can better understand your archiving requirements. Storage resource management (SRM) tools present the best approach for capturing this baseline. SRM tools can greatly improve budgeting and long-term planning, which in turn can offer significant ROI.

By implementing and using a storage resource management (SRM) solution, you’ll have a more accurate perspective of your storage infrastructure. SRM gives you accurate proactive capacity planning and more accurate growth forecasting. And with the storage performance trending and near real-time statistics, you have much better insight into your day-to-day data usage patterns. This provides early warning signs to potential performance and capacity-related problems and downtime.

In our next data archiving post, 2 of 5, we’ll be exploring a data archiving strategy for email. Be sure to chime in with your thoughts and come back soon for the next post.