5.4 KiB
Storage management a weak area for most enterprises
Survey finds companies are adopting technology for such things as AI, machine learning, edge computing and IoT, but still use legacy storage that can't handle those workloads.
Stop me if you’ve heard this before: Companies are racing to a new technological paradigm but are using yesterday’s tech to do it.
I know. Shocking.
A survey of more than 300 storage professionals by storage vendor NGD Systems found only 11% of the companies they talked to would give themselves an “A” grade for their compute and storage capabilities.
Why? The chief reason given is that while enterprises are rapidly deploying technologies for edge networks, real-time analytics, machine learning, and internet of things (IoT) projects, they are still using legacy storage solutions that are not designed for such data-intensive workloads. More than half — 54% — said their processing of edge applications is a bottleneck, and they want faster and more intelligent storage solutions.
[ Read also: What is NVMe, and how is it changing enterprise storage ]
NVMe SSD use increases, but doesn't solve all needs
It’s not all bad news. The study, entitled "The State of Storage and Edge Computing" and conducted by Dimensional Research, found 60% of storage professionals are using NVMe SSDs to speed up the processing of large data sets being generated at the edge.
However, this has not solved their needs. As artificial intelligence (AI) and other data-intensive deployments increase, data needs to be moved over increasingly longer distances, which causes network bottlenecks and delays analytic results. And edge computing systems tend to have a smaller footprint than a traditional data center, so they are performance constrained.
The solution is to process the data where it is ingested, in this case, the edge device. Separate the wheat from the chafe and only send relevant data upstream to a data center to be processed. This is called computational storage, processing data where it is stored rather than moving it around.
According to the survey, 89% of respondents said they expect real value from computational storage. Conveniently, NGD is a vendor of computational storage systems. So, yes, this is a self-serving finding. This happens a lot. That doesn’t mean they don’t have a valid point, though. Processing the data where it lies is the point of edge computing.
Among the survey’s findings:
- 55% use edge computing
- 71% use edge computing for real-time analytics
- 61% said the cost of traditional storage solutions continues to plague their applications
- 57% said faster access to storage would improve their compute abilities
The study also found that NVMe is being adopted very quickly but is being hampered by price.
- 86% expect storage’s future to rely on NVMe SSDs
- 60% use NVMe SSDs in their work environments
- 63% said NVMe SSDs helped with superior storage speed
- 67% reported budget and cost as issues preventing the use of NVMe SSDs
That last finding is why so many enterprises are hampered in their work. For whatever reason they are using old storage systems rather than new NVMe systems, and it hurts them.
GPUs won't improve workload performance
One interesting finding: 70% of respondents said they are using GPUs to help improve workload performance, but NGD said those are no good.
“We were not surprised to find that while more than half of respondents are actively using edge computing, more than 70% are using legacy GPUs, which will not reduce the network bandwidth, power and footprint necessary to analyze mass data-sets in real time,” said Nader Salessi, CEO and founder of NGD Systems, in a statement.
That’s because GPUs lend themselves well to repetitive tasks and parallel processing jobs, while computational storage is very much a serial processing job, with the task constantly changing. So while some processing jobs will benefit from a GPU, a good number will not and the GPU is essentially wasted.
Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
作者：Andy Patrizio 选题：lujun9972 译者：译者ID 校对：校对者ID