The cloud is a vast resource of compute and storage, all available on-demand as the organization needs it. Using cloud storage as VMware primary storage with compute remaining on-premises opens up many possibilities. In our last column we discussed how to solve the cloud’s biggest shortcoming, latency. With the “how” out of the way, it is time to address the “why.”
Reduced Storage Cost
On-premises storage can be fairly inexpensive, but even if it IT can buy it for the same price as cloud storage it has one big problem; the cost to upgrade it. Upgrades, also known as storage refreshes, require a complete swap out of at least the storage controller and very often the entire storage system.
Even scale-out systems struggle with upgrades. The addition of nodes to these systems typically come with both compute and storage, but in many cases the data center only needs to upgrade one of those resources. The other problem is when nodes are added they are typically added “full”, meaning they are maxed out in CPU and capacity. That means when the node is initially added there, is an excess of compute and capacity, which goes wasted.
When a VMware environment uses cloud storage as its primary storage target the organization can purchase storage capacity as it needs it and as granularly as it wants – terabytes at a time if need be. As capacity is added, there is no additional burden placed on the organization to come up with square footage to house the capacity nor is there an increase in the organization’s energy consumption.
When the cloud is used as primary storage, performance scales separately. The on-premises cache is the device that responds to most of the IO requests the organizations makes. If the organization needs better IO performance, it can upgrade the on-premises cache independent of any capacity decisions in the cloud.
Eliminating Copy Data
Copy data, the legitimate, purposeful copies of data that an organization makes, is such a big problem that a whole sub-category of solutions have appeared on the market to address it. Without copy data management the company has to manually make a copy of production data for various reasons. There is the obvious need for copy data for backup and disaster recovery but there is also the need to make copies of data for test-dev, DevOps, analytics, reporting, etc.
The manual copying of this data, making sure it is positioned for one of the above use cases and then, just as important, making sure that these extra copies are removed when they are no longer needed is a costly operation expense. In fact the cost of manual implementation and orchestration can far outweigh the physical cost of these copies.
Copy data management solutions help with some copy data problems. First, the solutions put a stop to extraneous copies. And in some cases they also automate and orchestrate the creation and placement of secondary copies. But they do have to make at least two extra copies of data, an on-site copy and an off-site copy. As a result, in addition to the primary copy, there are three copies of data, even with the best copy data management solution.
In both the manual and copy data management use cases the organization still needs to purchase additional compute resources for those processes. There is an obvious hard cost associated because of the purchase of new servers and increasing administration costs to manage the servers.
Using the cloud as primary VMware storage, again, helps alleviate the copy data problem. All data is in the cloud and is only cached on-premises. Data in the cloud is, of course, replicated to an additional location for disaster protection. But beyond that “copies” can be made in similar fashion to copy data management solutions by leveraging snapshots.
More importantly, these snapshots can be acted upon by cloud compute resources, the organization does not need to buy additional resources. Using cloud compute resources is almost a perfect match, since most process that need copy data only need it for a short period of time. Once the answer is achieved the environment could easily be collapsed. The problem is, if physical hardware that is involved collapsing is essentially impossible. With the cloud, it is very easy.
VMware storage has been a thorn in the side of VMware administrators from the moment it rolled into production. It’s no surprise since storage was a thorn in the side of IT long before VMware appeared. Using the cloud as primary storage for VMware environment resolves a lot of these issues and opens up new possibilities.
In our next column we’ll discuss the data protection and disaster recovery advantages of using the cloud as primary VMware storage.