In every cycle of technical innovation that we go through on IT infrastructure, I have witnessed a similar pattern. A gradual realisation that your strategic infrastructure decisions always end up around “where the data is”. Surprisingly, it can take people some time to realise this inevitable fact.
This first dawned on me years ago in the age of client-server computing back in the late 1980s. I was listening to a presentation from one of the best data storage consultants I have ever had the pleasure to work with, and he made this observation. Whenever he spoke to customers about a data centre refresh, they would always start by asking him about the servers. However, he always made the point that a data centre refresh needed to start with storage. His logic was that you can’t upgrade your servers without upgrading your storage, but you can upgrade your storage without upgrading your servers.
Now that we are in a time of the cloud-first enterprise, the same holds true. Decisions about cloud infrastructure need to revolve around where the data and storage need to be.
There’s a number of variables that affect the choice of cloud or on-premises for your enterprise storage, including cost analysis, lock in over time and data security. Every business will utilise cloud within its storage strategy, but there are some clear-cut use cases where on-premises will be a pre-requisite.
Storage has to meet the laws of physics.
Certain applications need to perform fast and at scale, and while cloud can do a lot of amazing things, it can’t rewrite the laws of physics. Namely, that data still has to travel between users, compute and storage. If the storage’s location is geographically remote from the users and compute, then it has to cover a large distance and this takes time, or to phrase it another way, “latency”.
As the digital revolution has taken hold, it has happened in tandem with the adoption of cloud and the proliferation of data-driven business. They go hand in hand but not all the time. Moreover, to a large extent, due to the laws of physics which dictate that distance = latency, there are two clear-cut situations where on-premises storage is known to be “the way to go”.
Business Critical Apps
Any business or mission-critical application that needs to process and act on data analysis at scale and split-second speed has to have storage on-premises, or near to the users and compute. High performance computing, video-intensive applications and even VDI are all examples of applications that may hit bottlenecks without having high performance storage on-premises. Companies like Huawei continue to invest in building high performance storage like the OceanStor Dorado V6 to meet the demanding needs of these data-intensive applications.
When analyst firm ESG tested Huawei’s OceanStor Dorado V6 with some of the most demanding mission-critical workloads such as data analytics or OLTP, they found that it was able to produce one of the lowest response times they had ever seen for that level of performance, from any vendor they tested.
Besides delivering on both the high IOPS and extremely low latency fronts, it’s also equipped with the HyperCDP function that was able to take continuous snapshots with no impact on performance, enabling point-in-time recoverability and ensuring that the critical data is well protected.
Edge computing is arguably the manifestation that even in the age of cloud, there are increasing use cases that do not lend themselves to cloud storage. The edge is all about keeping data and storage geographically close. For example, modern factories may have many thousands or even hundreds of thousands of IoT monitoring devices that send data for real-time analysis, leading to instant actions. The storage for collecting and analysing this data is increasingly located at the factory itself (or the edge) and connected to cloud or the primary data centre. The key is that storage needs to be placed on-premises at the edge.
For most companies, storage strategy will involve a combination of cloud and on-premises at their primary data centre and the edge. Storage strategy must revolve around the data itself, and this has become even more true in the cloud era. The variety of sources of data has increased, the volume has scaled, and the locations at which storage can be positioned have expanded. Understanding these dynamics is critical for any company that wants to develop a modern storage strategy.