6 minute read7 Tips for Highly Successful Storage Designs
By Dan Barber
Does your business have a five-year, or even a three-year, plan for storage? Can you, for example, project how much data you will need to store and serve for 2015 while also predicting performance levels to meet your business goals?
Sadly, many IT professionals do not practice “storage design with the future in mind.” That said, such a practice is key to not getting caught with your storage down, so to speak. More often than not, a piecemeal approach is taken whereby a company adds some storage here, a shelf there, and never moves from reactive to proactive. Is this because of a lack of resources, or is it a planning issue? Check out the points below and decide for yourself.
1. Preference: Vendor Capability or Vendor Consistency?
Question: Should I buy the system that best fits my needs, or should I find a solution that fits my needs and matches my current storage platform? The answer? It depends. As an IT Director or CIO/CTO, you have to make a decision that always pits homogeneity against flexibility. What is most important to you? Ease of management? The best feature set? Only you can decide. But once you decide, you should build your datacenter storage around that practice to create a consistent storage environment.
2. Unplanned Growth: Plan for It
The amount of data stored is increasing in orders of magnitude very quickly. Think about it: high-resolution graphics documents, once limited to the creative elite, now live on just about every personal computer world-wide. The pixel proliferation of digital photos and illustrations only compounds the matter. Data size will continue to increase even if your business technology remains status quo.
3. Requirements: Prepare a Detailed, Current Document
Regardless of what you have done in the past, it is vital to a storage design to gather requirements. The old maxim, “Well begun is half done” is applicable here. If you begin with a poor understanding of your own requirements, you will inevitably buy hardware that doesn’t meet your needs, and you will end up frustrated. It happens to businesses every day.
What does a good document look like with regard to requirements? Two examples.
- “Require 40,000 IOPS and 60TB of space.”
Now, this might seem like a good requirement while, in actuality, it is not. Do all 60TB need to function at this 40k IOPS level? Possibly. But the requirement is vague. A storage vendor could provide 20 100GB SSDs to meet the IOPS and then 20 3TB 7.2k disks to meet capacity. Do you think that will run your Exchange environment well? What about RAID protection?
A better requirement might be something like this:
- “Require 10TB usable of Tier 1 storage at 10,000 IOPS”
- “Require 40TB usable of Tier 3 storage for backup at 5,000 IOPS”
- “Require 10TB usable of Tier 0/1 storage for SAP at 25,000 IOPS”
Don’t forget to add requirements for uptime (how many 9s of availability do you require?) and redundancy, among other things.
4. Product/Service Growth: Plan for It
The problem with many projects is that by the time they are completed they have already been outgrown (think road construction). The same can easily happen to businesses in a tight economy, where it often takes several months or longer to secure funding for a tech refresh or upgrade. When planning your storage design, be sure to take into account forecasted business growth and product launches.
5. Obsolete Storage: What to do?
When the life-cycle of a product is complete, what should you do with it? That’s a great question. Generally, there are a few options. First, you could recoup some kind of cost either through a third-party buyer or used equipment. You may also consider a trade-up with the manufacturer. Or perhaps you could re-purpose the system from production to less a critical role, such as a DR site, a replication target, a backup target, or a testing and development environment. There are numerous ways to make older storage work in your environment.
6. Single Points of Failure: Eliminate Them
Which is top priority? Redundancy, capacity, scalability, reliability? What about performance? Capacities can be stretched; that is, scalability can increase and performance can be tuned. Redundancy, however, is an all-or-nothing game. Either you have redundancy, or you have a Single Point of Failure (SPOF). SPOFs should be eliminated for all production systems, whatever your IT policies deem those systems to be.
7. Embrace Professional Services
IT Managers often prefer their own personnel to set up their systems, and for good reason. They need to familiarize themselves with the products they will be managing. However, that does not necessitate a complete rejection of professional services.
Professional services can be all encompassing to be sure, which is how most IT folks think about them. Or, on the other hand, they can be tailored to suit your needs. Consider engaging technical services to verify your setup, perform load and failover testing, or perform some other kind of validation. A small investment can yield huge returns for your business.