Alibaba Cloud offered a peek at its latest homegrown silicon at its annual summit this week, which it calls Cloud Infrastructure Processing Units (CIPU).
The data processing units (DPUs), which we're told have already been deployed in a “handful” of the Chinese giant’s datacenters, offload virtualization functions associated with storage, networking, and security from the host CPU cores onto dedicated hardware.
“The rapid increase in data volume and scale, together with higher demand for lower latency, call for the creation of new tech infrastructure,” Alibaba Cloud Intelligence President Jeff Zhang said in a release.
However, the tech is hardly new, even for Alibaba Cloud. SmartNICs, IPUs, DPUs (call them what you want) have been knocking around hyperscale and cloud datacenters for years now. Alibaba's CIPU appears to be an evolution of the company's X-Dragon smartNIC, designed to compete head on with Amazon Web Services’ Nitro cards.
The exact architecture underpinning Alibaba’s CIPUs is unclear, however, the cards do use a standard PCIe card form factor.
Alibaba claims the accelerator is capable of reducing network latency to as little as 5 microseconds, while improving compute performance in data-intensive AI and big-data Spark deployments by as much as 30 percent, according to the company’s internal benchmarks.
The Register reached out to Alibaba Cloud for more details; we’ll let you know if we hear back.
DPUs have become a hot topic over the past few years as we’ve seen an influx of products from the likes of Intel, Marvell, Fungible, Nvidia, and AMD’s Xilinx and Pensando business units to name just a few.
All of these devices share a common goal: accelerate input/output intensive workloads — common in networking, storage, and security applications — by offloading them to specialized domain-specific accelerators, freeing CPU resources to run tenant workloads in the process.
In this regard, Alibaba’s CIPUs are nothing new, but still notable as being developed in house as opposed to using third-party smartNICs as Google Cloud Platform and others have done.
Alibaba’s CIPU comes just months after the cloud provider detailed an in-house microprocessor.
Developed by the company’s T-Head development branch, the Yitian 710 is based on a 5-nanometer manufacturing process, boasts 128 Armv9-compatible CPU cores clocked at 3.2 GHz, and supports the latest DDR5 and PCIe 5.0 standards.
According to Alibaba’s internal benchmarks, the chip is 20 percent more powerful and 50 percent more efficient than the current crop of server processors on the market.
Customers can deploy workloads on the chips now on Alibaba’s Elastic Compute Service (ECS) g8m instances.
Alibaba’s efforts closely mirror those taken by American rival Amazon, which was among the first to pursue custom silicon as a differentiator for its public cloud.
AWS offers a full suite of instances using any combination of its Graviton CPUs, Nitro smartNICs, and Trainium and Inferentia AI processors.
Alibaba is hardly the only cloud vendor now warming up to idea of custom cloud infrastructure. Microsoft is actively deploying Ampere’s Arm-based Altra CPUs in Azure and is rumored to be working on a custom processor of its own. ®
Updated Amazon has blasted a proposed antitrust law that aims to clamp down on anti-competitive practices by Big Tech.
The American Innovation and Choice Online Act (AICOA) led by Senators Amy Klobuchar (D-MN) and House Representative David Cicilline (D-RI) is a bipartisan bill, with Democrat and Republican support in the Senate and House. It is still making its way through Congress.
The bill [PDF] prohibits certain "online platforms" from unfairly promoting their own products and services in a way that prevents or hampers third-party businesses in competing. Said platforms with 50 million-plus active monthly users in the US or 100,000-plus US business users, and either $550 billion-plus in annual sales or market cap or a billion-plus worldwide users, that act as a "critical trading partner" for suppliers would be affected.
Google Cloud Platform (GCP) roped the Lone Star State into its cloud empire this week with the launch of its Dallas, Texas region.
The $600 million datacenter campus, which broke ground in 2019, is located approximately 25 miles south of the Dallas metro. The site, Google’s 11th in the US and 34th globally, is the latest in an ongoing effort to expand the cloud provider’s reach to new markets.
“We’ve heard from many of you that the availability of your workloads and business continuity are increasingly top priorities," Stacy Trackey Meagher managing director for Google’s central region, said in a statement. “The Dallas region gives you added capacity and the flexibility to distribute your workloads across the US.”
Google Cloud's Anthos on-prem platform is getting a new home under the search giant’s recently announced Google Distributed Cloud (GDC) portfolio, where it will live on as a software-based competitor to AWS Outposts and Microsoft Azure Stack.
Introduced last fall, GDC enables customers to deploy managed servers and software in private datacenters and at communication service provider or on the edge.
Its latest update sees Google reposition Anthos on-prem, introduced back in 2020, as the bring-your-own-server edition of GDC. Using the service, customers can extend Google Cloud-style management and services to applications running on-prem.
The major hyperscalers and cloud providers are forecast to spend 25 percent more on datacenter infrastructure this year to $18 billion following record investments in the opening three months of 2022.
This is according to Dell’Oro Group research, which found new cloud deployments and higher per-unit infrastructure costs underpinned capex spending in Q1, which grew at its fastest pace in nearly three years, the report found.
Datacenter spending is expected to receive an additional boost later this year as the top four cloud providers expand their services to as many as 30 new regions and memory prices trend upward ahead of Intel and AMD’s next-gen processor families, Dell’Oro analyst Baron Fung told The Register
Why build a cloud datacenter yourself, when you can rent one from Hewlett Packard Enterprise? It may seem unorthodox, but That’s exactly the approach Singapore-based private cloud provider Taeknizon is using to extend its private cloud offering to the United Arab Emirates (UAE).
Founded in 2012, Taeknizon offers a menagerie of services ranging from IoT, robotics, and AI to colocation and private cloud services, primarily in the Middle East and Asia. The company’s latest expansion in the UAE will see it lean on HPE GreenLake’s anything-as-a-service (XaaS) platform to meet growing demand from small-to-midsize enterprises for cloud services in the region.
“Today, 94% of companies operating in the UAE are SMEs," Ahmad AlKhallafi, UAE managing director at HPE, said in a statement. "Taeknizon’s as-a-service model caters to the requirements of SMEs and aligns with our vision to empower youth and the local startup community.”
Updated The US House Oversight Committee has told Amazon CEO Andy Jassy to turn over documents pertaining to the collapse of an Amazon warehouse – and if he doesn't, the lawmakers say they will be forced to "consider alternative measures."
Penned by Oversight Committee members Alexandria Ocasio-Cortez (D-NY), Cori Bush (D-MO) and committee chairwoman Carolyn B. Maloney (D-NY), the letter refers to the destruction of an Edwardsville, Illinois, Amazon fulfillment center in which six people were killed when a tornado hit. It was reported that the facility received two weather warnings about 20 minutes before the tornado struck at 8.27pm on December 10; most staff had headed to a shelter, some to an area where there were no windows but was hard hit by the storm.
In late March, the Oversight Committee sent a letter to Jassy with a mid-April deadline to hand over a variety of documents, including disaster policies and procedures, communication between managers, employees and contractors, and internal discussion of the tornado and its aftermath.
Amazon's attempt to dismiss a lawsuit, brought by one of its senior software engineers, asking it to reimburse workers for internet and electricity costs racked up while working from home in the pandemic, has been rejected by a California judge.
David George Williams sued his employer for refusing to foot his monthly home office expenses, claiming Amazon is violating California's labor laws. The state's Labor Code section 2802 states: "An employer shall indemnify his or her employee for all necessary expenditures or losses incurred by the employee in direct consequence of the discharge of his or her duties, or of his or her obedience to the directions of the employer."
Williams reckons Amazon should not only be paying for its techies' home internet and electricity, but also for any other expenses related to their ad-hoc home office space during the pandemic. Williams sued the cloud giant on behalf of himself and over 4,000 workers employed in California across 12 locations, arguing these costs will range from $50 to $100 per month during the time they were told to stay away from corporate campuses as the coronavirus spread.
Alibaba's cloud business and financial services affiliate Ant Group has expanded further out of China this week, by opening a pair of datacenters in Saudia Arabia and a digital wholesale bank in Singapore.
Alibaba Cloud and Saudi Telecom Company (STC) have opened two cloud services in Riyadh which will serve as a regional hub as part of a joint venture called the Saudi Cloud Computing Company (SCCC). STC confirmed the launch on Tuesday and the joint venture, SCCC, shared scenes from the launch.
Other businesses playing a part in SCCC are eWTP Arabia Capital, the Saudi Company for Artificial Intelligence (SCAI), and the Saudi Information Technology Company (SITE).
AWS is trying to help organizations migrate their mainframe-based workloads to the cloud and potentially transform them into modern cloud-native services.
The Mainframe Modernization initiative was unveiled at the cloud giant's Re:Invent conference at the end of last year, where CEO Adam Selipsky claimed that "customers are trying to get off their mainframes as fast as they can."
Whether this is based in reality or not, AWS concedes that such a migration will inevitably involve the customer going through a lengthy and complex process that requires multiple steps to discover, assess, test, and operate the new workload environments.
Concern is growing that a World Trade Organization (WTO) moratorium on cross-border tariffs covering data may not be extended, which would hit e-commerce if countries decide to introduce such tariffs.
Representatives of the WTO's 164 members are meeting in Geneva as part of a multi-day ministerial conference. June 15 was to be the final day but the trade organization today confirmed it is being extended until June 16, to facilitate outcomes on the main issues under discussion.
The current moratorium covering e-commerce tariffs was introduced in 1998, and so far the WTO has extended it at such meetings, which typically take place every two years.
The Register - Independent news and views for the tech community. Part of Situation Publishing
Biting the hand that feeds IT © 1998–2022