Optimizing IT: How much can you save on storage?

I meet with the IT staff and upper management of large organizations all the time, and I am continually amazed at how almost everyone is suffering under similar problems.

It’s getting to the point that I can predict their problems by asking which vendors they use.

Common IT problems:

• Backup windows
• Network bandwidth
• Tape media and backup process
• Low storage utilization
• Storage resource management
• Problems with service levels
• Data migration issues
• Disaster recovery complexity and cost
• Operational complexity and costs
• Security and risk mitigation

What I find most interesting is the fact that many organizations (especially those without visionary leadership or internal technical engineering standards) are still stuck with dated operational processes and technologies, which are typically restricted to their current vendors' technical capabilities and supported operational methods.

Traditional IT methods focus on solving individual issues with specific budgets and vendor solutions, while applying management models based on standards like Information technology infrastructure library (ITIL) to keep costs down.

Examples of traditional IT methods:

• Array-based replication without WAN optimization
• LAN-based tape backup
• Offsite tape shipping
• Physical servers for disaster recovery
• Internally controlled DR process to secondary data center
• Multi-vendor disk tiers as an attempt to lower costs and provide manual information lifecycle management
• Multiple service levels and methods for application protection and recovery

The new method of optimizing IT takes a more holistic view of the problem, and leverages innovations in technologies and services to increase efficiency and get better return on existing assets. This in turn enables IT management to combine the budgets of multiple projects to provide better purchasing power, and through innovations such as virtualization or cloud computing, provide data services at commodity prices.

I have seen some companies simplify their operations and achieve some palpable improvements in expenses through the use of recent innovations in technology such as storage virtualization, data deduplication, continuous data protection, virtual tape, cloud computing, etc,

Example expense reduction after IT optimization:

• 330% improvement in production storage costs
• 680% improvement in backup costs
• 1200% improvement in outage costs

Don't just take my word for it. You can ask my friends over at Hitachi Hu Yoshida or David Merrill how inefficient many organizations really are just in storage utilization alone, and how just adding virtualization with thin provisining can reduce costs dramatically.

The sad part is that due to all the complexity in existing IT operations, and what I can only assume may be intellectual laziness on the part of some IT management, some organizations have relegated their day-to-day operational responsibilities to outside consultants, or just let their vendors run their shops.

This would not be a problem if they were offsetting their costs by outsourcing to a cloud provider for certain functions, but most are just so dependent on their vendors (which I am sure their vendors love), they are reluctant to change, and therefore are not reaping the benefits of recent innovations that could lower their costs. In other words, their vendors have them locked in.

Even now, in many IT shops, storage utilization rates are still abysmal. Few are using more than 40 percent of what they purchased. Implementing a storage area network (SAN) was supposed to solve that problem, but many IT shops still have over-provisioned islands of storage which cannot be shared with other applications. Some of this stranded unused storage is dedicated to logical storage units (LUNS) that were over provisioned, and locked in for exclusive access to existing databases.

Without thin provisioning or storage virtualization, the storage in those LUNS cannot easily be recaptured and reallocated as required by the business. Other stranded, unused storage may be dedicated as network attached storage (NAS), and since it's usually implemented apart from the SAN, it cannot be migrated back to the SAN for other applications when it's underutilized.

Adding SAN or NAS storage resources is typically a manual time consuming process. If you don't plan and allocate storage perfectly based on what is actually required by each application, storage resources get underutilized and wasted, which increases costs. This is why the art of capacity planning a currently a must for the organization.

The figure below represents a traditional datacenter model where storage is allocated from different storage tiers.

In order to solve some of the cost issues, many IT shops are creating tiers of storage. They place critical applications on the expensive fast disks, and the lesser applications on cheaper disk. The different storage tiers are purchased either from a single preferred vendor, or to save money, end users pit vendors against each other and purchase from the vendor coming in with the lowest price.

Both procurement approaches end up costing more than they should. Sticking with a single vendor locks you in, and although pitting vendors against each other may provide savings of up front capital costs, you end up paying in the long run via higher operational costs.

Without innovations like storage virtualization, moving data between tiers or migrating between different arrays is usually a manual process, so operational costs stay high.

All of these traditional approaches add up to higher costs. As an example, let's map out what the storage costs of a typical data center using a traditional model.

Example per GB storage cost model for traditional tiers:

• Tier1 with BCV: $3.00 RPO = 12 Hours
• Tier1 with DR: $6.00 RPO = 12 hours
• Tier2 with BCV: $2.50 RPO = 24 Hours
• Tier2 with DR: $3.50 RPO = 24 hours
• Tier3 with BCV: $2.00 RPO = 24 Hours
• Tier4 SAN: $1.00 RPO = 24 Hours
• Tier5 Tape w/offsite: $0.85 RPO = 24 Hours

Now let's do some math. To keep this simple, I will use 20TB as the current SAN capacity (skipping the NAS storage to keep the math simple), with 2 percent data growth and 3 percent data change spread evenly across all storage tiers. Critical data is 3TB on T1 and T2 storage.

For backup, let's use a two-month data retention policy using standard backup software, and a policy of daily incremental and weekly full backups. Typical array-based replication will be used for the three TB of critical apps for DR for T1 and T2 (the other 2TB of production data is local BCV only) and offsite tape for DR of other apps across tiers (14TB). We will assume a $1000.00 per hour downtime cost for this company, and the company owns it's own DR location.

Let's also assume all other costs (tape recall, operating expense, manpower, hardware and software, etc.) are built into the monthly per GB prices for each tier.

Current Capacity: 20TB across 4 tiers, plus T5 backup

• T1 =2TB BCV + 3TB DR
• T2 =2TB BCV + 3TB DR
• T3 = 5TB
• T4 = 5TB
• T5 = 101TB of tape (20TB + (growth + change) x retention)

Current Monthly Expenses: $56,320 for production and $87,910 for backup

T1 = $24,576 / month

T2 = $15,872 / month

T3 = $10,240 / month

T4 = $5,632 / month

T5 = $87,910 /month

Single outage costs: $12K for T1-T2 and $24K for T3-T5 (at $1,000 per hour)

As you can see, backup and DR are always the big budget busters, and since they add the least value to the business (until someone loses data), they are where most organizations skimp on budget. So how do you lower your costs, especially for backup and DR?

In the next blog, I will discuss how you can leverage innovative technologies to reduce your IT spending while improving operations and service levels.

Christopher Poelker is the author of Storage Area Networks for Dummies, and he is currently the vice president of Enterprise Solutions at FalconStor Software.