Vertical integration in the storage world is more often than not a repackaging effort more that it is a tight integration between previously separate products. As we point out in our recent article "The Storage Challenges to Oracle Exadata," vendors often combine products from multiple companies that they have purchased and then claim that they are vertically integrated. At least in Oracle's case they did add some specific software capabilities to Exadata that allows the Oracle software to uniquely take advantage of the underlying Oracle hardware.
What is real vertical integration?
First, what is not vertical integration? Shipping all the products on the same pallet is not vertical integration. But we have seen some vendors do that and claim it to be so. Making sure that all the logos match is not vertical integration. Finally, vertical integration is not even making sure that a particular set of servers, software, and hardware are tested to work together.
The first two cases are examples of vendors stretching their marketing freedoms a little too far. The last example, testing configurations, has value--we call this building a reference architecture. We have seen NetApp do this with their FlexPod initiative, but they correctly point to this as a reference design not a vertically integrated stack. They fully expect every configuration to be slightly different. Our independent checks of the channel and end users indicate that this strategy is proving successful. We have seen other vendors take a more inflexible approach and again, according to the news and our checks, they are not being as successful.
With these reference architectures or pre-bundled configurations, many vendors will claim vertical integration because they add some software to the equation, like a centralized management GUI or again, in Oracle's case, additional tweaks to their software for better performance. These are good and worthwhile additions, but they do not equal vertical integration.
Vertical integration then, especially when you claim a comparison to Apple, is not simply making sure it all works together and adding a couple of extra features to your software. It is building, from the ground up, a completely integrated system that can't be recreated from standalone software and hardware.
The big question though, is vertical integration or as we have seen thus far pre-bundled solutions, good for the storage industry? The answer, as always, depends on what is important to the IT professional responsible for day to day operations. If the integration of the components works well together and delivers an easier installation and more efficient operations without too much compromise on performance, then the integrated stack may be something to consider.
If flexibility, affordability, and performance are of greater concern then a non-integrated approach may be better for you. We have seen several multi-vendor configurations claim better performance results than allegedly vertically integrated solutions and do so at a significantly better price point. The tradeoff may be a more complex integration and day to day operational process. How much weight you give that depends largely on the availability and skill of your internal IT staff and the abilities of the vendors involved in the solution.
Since I am mentioning vendors in this particular blog, it is important that I disclose a few relationships. As a point of disclosure, I am not against vertical integration as long as it is truly integration and not product bundling. For example, I have an iPhone/iPad and love the products. I am also not against product bundling. I just want vendors to call things what they are. Specifically, Oracle is not a client of my firm Storage Switzerland and NetApp is.
Follow Storage Switzerland on Twitter