If you're still stuck deciding which storage solution is best for you, or are sceptical about what the cloud can offer, I am here to reveal the truth behind some common misconceptions.
Yes, SSDs cost more than classic spinning disks to buy, I’m not denying that. However, this is the capital cost - instead look at the cost for performance. If you have a system that needs low latency at high IOPs then you need to purchase a vast disk based system to achieve this, with flash you get this within 2-4U and 24 disks typically.
If you look at it at face value £0.02 per GB is pretty cheap for storage. One of the top use cases for cloud is backup, so let’s look at the cost over time (sorry for the maths).
Let’s say you have 50TB of data that you want to backup, every year your data is growing by 20% and you want to retain daily incrementals for a week, weeks for a month and monthlies for a year.
For this backup policy we would normally expect 1.7 times the base amount of storage required to store the backups. This leaves us with the below costs:
Cost Formula: Base TB * (Growth ^ Number of Years) * Backup Factor * Conversion to GB * Cost of GB * Months
Year 1 Cost: 50*(1.2^0)*1.7*1024*£0.02*12 = £20,889.60
Year 2 Cost: 50*(1.2^1)*1.7*1024*£0.02*12 = £25,067.52
Year 3 Cost: 50*(1.2^2)*1.7*1024*£0.02*12 = £30,081.02
Year 4 Cost: 50*(1.2^3)*1.7*1024*£0.02*12 = £36,097.23
Year 5 Cost: 50*(1.2^4)*1.7*1024*£0.02*12 = £43,316.67
Total: £20,889.60 + £25,067.52 + £30,081.02 + £36,097.23 + £43,316.67 = £155,452.04
As you can see, the storage isn’t as cheap as it may first appear once you do the maths.
It’s commonly stated that backups need to be kept for five or seven years. The question I ask is why? Often the response is: “Finance say that we need to keep seven years of records”. That’s fine, but if you look into your finance system I can almost guarantee that you have at least seven years of records in your production system. So why are you keeping seven year old backups? All they are doing is costing you money.
According to research completed by Veritas, out of all of the unstructured data stored, over 50% of this is “dark” and unclassified. This data could be of value to the business, or it could have no value at all and is costing you money.
Based on the same research by Veritas, they identified ROT data, which is data that is redundant, obsolete or trivial. This ROT data accounts for 33% of all business data and provides no value. It could be multiple copies of a file held within a file server in different places, data that has no discernible owner, or Christmas party pictures from five years ago. All of these provide no value to a business.
If organisations are storing data that is redundant, obsolete or trivial, then why are they keeping it? Normally it is the fear of deletion. With GDPR coming it is going to become more common place for IT to delete this data as it lowers the remit that GDPR can apply to. If there is no business value, get rid of the data.
According to the Oxford English Dictionary, a backup is “a copy of a file or other item of data made in case the original is lost or damaged.” By this definition, a snapshot, a point in time version of a file or LUN, can’t be a backup.
If you look at this in a wider sense, in order for something to be classed as a backup it needs to be a copy of the data sat on a physically separate device. This could be a secondary array, the cloud, tape, or whatever you like. It is also worth noting that using a backup product and pointing at the same array you are backing up from is also not, by definition, a backup.
This is a classic RFP question, the required features are deduplication and compression. If you look at that it might seem like a reasonable request, but let's actually look at why all is not quite as it appears.
The main reason people want data reduction is to lower the cost of storage (there are some other minor benefits which for the sake of simplicity I am ignoring), this therefore doesn’t make it a feature. Instead you should be looking at the cost per effective TB:
If one array costs £100,000 and provides 10TB of storage with 10:1 data reduction, 100TB effective, it is costing you £1,000 per TB. If another array provides you 100TB of storage for £90,000, but no dedupe and compression, it is costing you £900 per TB.
If you look at this from a risk perspective, there is less risk with the second array as you aren't having to rely on achieving a high data reduction ratio for the platform to best cost effective. It is always worth getting data reduction ratios underwritten where possible, as this will avoid additional spend due to an underwhelming ratio.
To understand more about storage and data, why not join our upcoming Data Management with Ultima and Veritas event on Friday 23rd March. You will learn how an integrated enterprise data management solution can help you minimise risk and lower costs, whilst optimising your storage.
- By Matthew Beale (Storage Solutions Specialist)