IDEAS Insights

Dell’s 12G Servers Are Engineered to Take the Heat

Dell rolled out a new line of 12G servers this week that it claims are more powerful and much easier to manage than previous generations. Like similar announcements from competitors, this announcement focused on Dell’s high-level strategy and capabilities instead of detailing the components inside the boxes. For hardware, Dell is updating virtually all of its two-socket blades, towers, and rack servers with the new architecture. But it is not necessarily the core server hardware that catches our interest; it is all of the extras Dell has wrapped around the hardware. During development of the 12G servers, Dell says it received design input from 7,700 customers in 17 countries and four continents, and 500 evaluation units have been out since October, ensuring that Dell got this one right.

In the Race for Infrastructure Leadership, Software Trumps Hardware

This is the last of four articles that review the details behind IDEAS predictions for 2012 (see earlier postings on drivers of cloud computing adoption, the changing landscape of client devices, and the impact of solid-state-drive [SSD] storage).

In the cloud era, when users are increasingly shielded from details about underlying platforms, the challenge of differentiating server hardware has become more acute than ever. Like gravity, standardization is pulling IT infrastructure toward common hardware components. With every new generation, x86 servers and industry-standard operating systems such as Linux and Windows are being used to host increasingly critical workloads. To remain competitive, systems vendors are constantly under pressure to lower prices by cutting hardware and operating system margins. Faced with an identical "race to the bottom," vendors are stepping up efforts to complement their hardware with highly differentiated software that is optimized for their platforms.

Mellanox Poised to Attack the Storage Network Market

Mellanox is a proven solution enabler for high-performance computing (HPC). Its InfiniBand (IB)-based interconnects have been widely used in HPC clusters, including clustered storage systems. However, customers have often overlooked the fact that Mellanox is also an end-to-end fabric provider with high-quality, low-priced Ethernet-based products. Recently, Mellanox announced a new top-of-rack 10/40GbE switch with industry-leading density and industry-first PCIe 3.0-compliant network interface cards (NICs). These new products should help solidify Mellanox’s technical leadership in the highly competitive Ethernet fabric market. With more and more network storage systems offering support for iSCSI and Fibre Channel over Ethernet (FCoE) host connectivity, IDEAS expects Mellanox to play a more significant role in the storage network market, especially considering that Mellanox’s storage optimization technologies are already embedded in its network interconnect offerings.

Source: Mellanox

The Complexity of Measuring Cloud’s Impact on Costs

Despite the name, total cost of ownership (TCO) is not the measure of "total cost," as used by economists, but the measure of "accounting cost." TCO calculations measure some costs and are useful for business decisions, yet these calculations always leave out other costs that inevitably only the user can define. Analysts and end users alike will benefit from recognizing that TCO calculates only the accounting cost and does not include the opportunity cost, which includes the cost of the "road not taken," in a cost scenario. Opportunity cost needs to be included in a true analysis of total cost.

Figure 1 - Production Possibility Frontier Graphs Tradeoffs
(Above, a manufacturer with a limited amount of resources, which describes all real-world scenarios, can only produce guns or butter. The manufacturer can produce a mix of guns and butter, but there are tradeoffs in the amount they can produce of each.)

Widespread Adoption of Flash in 2012 Will Force Vendors to Improve Storage Management Software

This is the third of four articles that review the details behind IDEAS predictions for 2012. The first two articles focused on the drivers of cloud computing adoption and the changing landscape of client devices.

In 2012, the lines between servers and storage will continue to blur as systems vendors focus on solutions that integrate flash-based solid-state drives (SSDs) into servers, either complementing or replacing hard-disk drives (HDDs). As a result, a need will arise for storage management software that can bridge storage arrays and servers. Automated storage management solutions will also become increasingly important in emerging markets.

Power Management Becomes Much Smarter in HP’s New Gen8 Servers

After accidentally leaking details of its eighth-generation ProLiant servers and blades last week, HP has finally launched the new servers at an event in Las Vegas. Gen8 servers cover the gamut, including rack servers, blades, towers, and scalable systems. This announcement was a little different than past announcements in that HP kept very quiet about the details of what is actually inside the servers. Instead, HP concentrated on high-level capabilities. Despite the lack of hard data, we learned that these servers have some very impressive capabilities that will definitely help datacenter managers deal with some major problems they face today.

First Look at HP ProLiant Gen8 Servers

The engineering behind HP’s new generation of ProLiant servers appears to be an effort to cut down the amount of time spent on diagnosing, maintaining, and repairing servers. After analyzing tens of thousands of support cases on prior-generation servers, HP has sought in ProLiant Gen8 to address the issue of time-to-repair head on. According to Jeff Carlat, director of ISS software at HP, the new capabilities built into the Gen8 servers will let HP “reduce the total time to problem resolution by 66%.”

Where to Put Flash for Enterprise Performance?

The approach to utilizing flash for high-performance, enterprise use cases is evolving. Compute-intensive servers have already been enlisted to enable high-performance applications like enterprise resource planning (ERP), customer relationship management (CRM), online transaction processing (OLTP) databases, and more recently virtual desktop infrastructure (VDI). But a high-performance storage architecture to match the high performance compute has not yet been decided on. Flash seems to be the core of the solution, but where to put this flash for optimal enterprise performance? Announcements this week from EMC and IBM suggest two approaches: one server-based, one array-based.

Amazon’s Cloud Already a Billion Dollar Business?

By now, most technology industry analysts have released their predictions for 2012. Many, including IDEAS, forecast that Amazon AWS would become the first cloud venture to have over $1 billion in revenue. We were all probably wrong.

That's because Amazon AWS likely already reached that mark in 2011. The financial results for its cloud business are reported by Amazon in a category called "other," which Amazon defines as "AWS, miscellaneous marketing and promotional agreements, other seller sites, and co-branded credit card agreements." Revenue from "other" for the past decade is shown in the chart below, which is sourced from Amazon's own quarterly report. However, by separating the sales specific to Amazon's cloud business from "other," we can get a pretty good idea of Amazon's revenue from cloud.

 Figure 1. Amazon Sales for "Other" Category. Source: Amazon Quarterly Reports

What Server Power Calculators Can (and Can’t) Do For You - Part 2 of 2

In my prior blog post, I compared the capabilities offered by server vendors in their power calculator tools. In this post, I'll give my recommendations for how to use these tools, talk about their accuracy, and suggest some improvements.