SADA, Intel Piloting Cloud Optimization Program For Google Cloud Customers
‘When you’re running big applications and you’re going to consume thousands to tens of thousands of computers, the performance differences between processors become really important,’ said Miles Ward, chief technology officer at SADA, a Los Angeles-based business and technology consultancy.
SADA is piloting a program with chipmaker Intel to help Google Cloud customers optimize their hybrid and multi-cloud workloads and compute spending.
The new cloud optimization program is aimed at ensuring those organizations are using the right Intel technology in their Google Cloud Platform (GCP) environments to perform as efficiently and cost-effectively as possible, according to SADA, a Los Angeles-based business and technology consultancy.
“When you’re running big applications, and you’re going to consume thousands to tens of thousands of computers, the performance differences between processors become really important,” said Miles Ward, chief technology officer at SADA, a Google Cloud Premier Partner and 2020 Google Cloud Reseller Partner of the Year for North America.
“There is variability not only between the different kinds of processors across the different manufacturers, but there’s a lot that you can do in really minor tuning changes to the way software works and configurations that you use inside of the virtual machines or other tools to make a really big performance difference,” Ward told CRN. “And then you have this whole need with customers to be able to anticipate what the savings or what the performance benefit of moving to cloud will be. If you apply a little work, you can make things faster and cheaper.”
While SADA initially is offering the cloud optimization program to its larger customers, it may eventually make it available to all customers, according to the company. Currently, eligibility is based on customers meeting a level of annual spending on GCP and related Google Cloud services, using specific Intel hardware (N2, C2, M2, M1 or O2 instances) and running specific workloads such as smart analytics, data warehousing, infrastructure modernization or application modernization.
SADA vets and validates customers, making sure they have workloads that are going to benefit from the optimization it’s offering, such as big Hadoop clusters or large NoSQL and relational database clusters – “places where computers are doing fairly predictable things fairly frequently at high volume,” according to Ward. Customers with more than $1 million in compute spend are those who’ll typically see a big enough performance impact to merit the time, he said.
“The results for customers have been really strong,” Ward said. “You can imagine in some of these cases, we’re talking about even single-digit percentage improvement in processing performance. If you’re running tens of thousands of virtual machines all the time, a couple percent is a big deal -- it’s thousands to hundreds of thousands of dollars, maybe millions of dollars in some cases. If you’re a small customer, 1 (percent) or 2 percent probably just doesn’t make that much of a difference.”
Participating customers, on average have reduced their compute costs by approximately 5 percent, according to Ward.
“By demonstrating the faster performance or the lower cost [Intel is] able to demonstrate not only that their systems for customers have gotten more efficient, but they are substantially more price-efficient than the offerings from their competitors,” Ward said. “That’s a big point Intel is trying to use this program for -- to make sure that they’re able to demonstrate how competitive they are in what is a very competitive market.”
Intel did not respond to a request for comment . In May, the semiconductor giant said it planned to introduce a new tool this year -- the Intel Cloud Optimizer -- to help partners optimize the performance and total cost of ownership of their workloads on cloud instances powered by its processors. The tool is part of Intel’s efforts to defend its dominant position in the cloud infrastructure market against chipmaker AMD and other companies.
SADA is not using that tool, but its cloud optimization program is part of that same Intel initiative, according to Ward. The partnership stemmed from brainstorming sessions with Santa Clara, Calif.-based Intel about its ability to give some of SADA’s larger customers the information, context and tuning control required to extract the maximum performance out of Intel’s most modern processor types.
“They’d see a big performance benefit, customers would see a big ROI [return on investment],” Ward said. “We’re a good clearinghouse. Because we’re the No. 1 resell partner for Google, we have access to some of the very biggest customers who benefit the most from those kinds of optimization. It’s a big story for customers to have access to Intel’s engineers directly that are doing this kind of tuning and performance optimization work.”
Among the dozen SADA companies already taking part in the program is Lilt, a modern language service and technology provider based in San Francisco that’s building a machine learning system that’s dependent on compute performance.
“Customers are looking for better translation quality, and Lilt, powered by Intel Xeon Scalable processors on Google Cloud, is able to delivery 2.5X better latency utilizing Intel optimization for TensorFlow and SADA professional services,” Thomas Zenkel, a research scientist at Lilt, said in a statement provided by SADA.