Fujitsu
Not logged in » Login
X

Please login

Please log in with your Fujitsu Partner Account.

Login


» Forgot password

Register now

If you do not have a Fujitsu Partner Account, please register for a new account.

» Register now
Oct 17 2016

Business Continuity Meets Hyper-Convergence: PRIMEFLEX Cluster-in-a-box

/data/www/ctec-live/application/public/media/images/blogimages/graphics/PF_Cib_Headshot.png

Over the course of its history, Fujitsu has always made a point of bringing the latest, most advanced ICT products to midmarket organizations and SMBs. Some decades back, this meant manufacturing and selling standardized telecommunications equipment. Today, we're helping the same class of customers build resilient, highly available, efficient, and easy-to-manage IT infrastructures – with solutions like our FUJITSU Integrated System PRIMEFLEX Cluster-in-a-box.

Upon its introduction in September 2012, Cluster-in-a-box was billed as the "fast track to continuous operation for SMBs." Since then, we have continually enhanced the package; the 2014 edition, for example, saw the addition of flexible configuration options. This time our focus is on hyper-convergence, i.e. delivering a miniature datacenter that includes all required components but saves up to 80% space compared with regular setups. Moreover, the new edition comes equipped with latest-generation processors and memory modules that effect substantial performance gains.

Hyper-Converged Infrastructures (HCI)
So what exactly is "hyper-convergence"? The term was originally coined by Arun Taneja, founder and chief analyst of consulting firm Taneja Group, who used it to describe a new class of systems that integrate compute, storage, networking and server virtualization capabilities under one hood. The entire system or "infrastructure" (hence the HCI moniker) may then be used for various purposes: in SMB environments and branch offices, it could run all existing workloads, whereas in large data centers it could serve as a building block that supports one or two key applications and works alongside other systems of similar type. Put differently, a hyper-converged system or HCI is basically a pre-configured appliance that can replace complete data center setups or larger chunks thereof, depending on the environment it is deployed to.

Parallels
The above description will seem somewhat familiar to long-term Fujitsu customers and channel partners, as it shares some characteristics with our PRIMEFLEX family of Integrated Systems and specifically Cluster-in-a-box. Like HCI, PRIMEFLEX systems combine complete sets of hardware and software components into single units that can be rolled out to data centers at any time and cover a wide range of usage scenarios from 'simple' server virtualization to Big Data analytics and high-performance computing. In this context, PRIMEFLEX Cluster-in-a-box was designed for organizations that didn't have the financial resources and/or skilled IT teams needed to build, manage and maintain regular server clusters, but still required typical 'cluster features' like transparent failover or straightforward disaster recovery in their daily operations. In other words, PRIMEFLEX systems in general and Cluster-in-a-box in particular were hyper-converged from the get-go, even before the term caught on as a marketing buzzword.

Improvements
Cluster-in-a-box 2016 is based on a single scale-out server model, the PRIMERGY CX400. That means, server functionalities are now the same across the board, and the entire solution is easier to administer. Contrary to what one might expect, switching to a single server platform will have no negative impact regarding freedom of choice – customers still can pick configurable systems or fixed-size appliances according to their individual needs:

  • The configurable version features two PRIMERGY CX2550 M2 server nodes, each of which is equipped with two 4th-generation Intel® Xeon® E5-2600 processors that hold between 6 and 18 cores and deliver clock speeds between 1,7 and 3.3 GHz. Main memory also received a notable upgrade – not only did we double the capacity from 256 to 512 GB of RAM, but we also replaced the classic DDR3 modules with fresher, faster DDR4 modules. In terms of internal storage, each configurable Cluster-in-a-box has room for 18 small form factor (2.5 inch) disks instead of 12. The total capacity will vary based on usage scenarios and the media deployed: if speed is paramount, users will likely opt for one of our "enterprise performance" SAS SSDs with 200, 400 or 800 GB of disk space; in contrast, organizations with large data repositories will prefer SAS HDDs that ingest 600, 900 or 1,200 GB of information and spin at 10,000 RPM. Since each Cluster-in-a-box has to include at least 3 disks, performance-oriented configurations start at 600 GB and max out at 14.4 TB, while their capacity-oriented counterparts provide no less than 1.8 TB to begin with and may reach a maximum of 21.6 TB. All disks connect to the server nodes via 12 Gb/s SAS Expanders. So altogether, the two-node standard edition of the configurable PRIMERGY Cluster-in-a-box offers better processors, more and faster RAM, 50% more disk space and 100% more internal bandwidth than the immediate predecessor – enough to handle up to 200 virtualized workloads or run in a physical cluster. However, all these detail improvements pale in comparison to the biggest one – which is that customers can now expand their configurable Cluster-in-a-box with two more nodes to reach a total of four nodes that run 8 top-of-the-line processors and feature 1 TB of RAM. For an overview of the improvements, please see the table below.
  • The upgrade for the fixed-size appliance was also substantial, although the results may not appear quite as impressive at first sight. Like the configurable version, it packs two PRIMERGY CX2550 M2 server nodes into a 2U enclosure. But because it was designed as an entry-level solution, the appliance will handle fewer virtualized workloads (up to 30) and should not be part of a physical cluster. These limitations notwithstanding, the hardware specs indicate that the system will meet average user needs: both server nodes are built around a duo of Intel® Xeon E5-2630 v4 processors with 10 cores and clock speeds ranging from 2.2 GHz in regular to 3.1 GHz in turbo mode. Main memory capacity was set to 128 GB for each node, and internal storage consists of eight 2.5 inch SAS HDDs with 1.2 TB capacity each, bringing the total allocated disk space to 9.6 TB. As in the above system, the disks connect to the hosts via 12 Gb/s SAS Expanders. Judging from this setup, it's probably fair to call the fixed-size appliance a "value-for-money edition" rather than an entry-level solution.

Another main advantage of PRIMEFLEX Cluster-in-a-box is that the servers come with a comprehensive, pre-installed software package including Windows Server 2012 R2. Customers who decide to buy the configurable edition can choose between licenses for Windows Server 2012 R2 Standard or Datacenter, each of which covers two processors. One of these licenses has Microsoft Cluster Services and Hyper-V activated by default; customers who wish to run a physical cluster have to turn off Hyper-V after the initial server configuration. The fixed-size appliance has Windows Server 2012 R2 Datacenter pre-installed; here the license covers two processors (one node) and the right to run unlimited virtual servers. Cluster Services and Hyper-V are turned on by default, and the installation includes four pre-installed, activated VMs running the host OS.

Image

Table 1: Comparison of PRIMEFLEX Cluster-in-a-box, configurable edition – previous and new version

As is the norm with PRIMERGY servers, each server node is equipped with Fujitsu's own, time-tested ServerView management suite. The package comprises helpful tools for on-site as well as remote administration; special attention has been paid to making the installation process for Cluster-in-a-box as easy as possible – all a user has to do is plug in the hardware and run our OOBE (Out-of-the-Box-Experience) Wizard and follow the onscreen instructions, and the setup will usually finish in 45 minutes or less. What's more, each node has the ServerView Agents pre-installed; this ensures easy integration into existing infrastructures.

Conclusion
So-called 'IT trends' often have their origins in marketing buzz, and that makes it hard to figure out which ones reflect genuine customer needs and which ones are based on pure PR. With regard to hyper-convergence, the distinction is easy: most SMBs and large numbers of midmarket organizations and enterprises go to great lengths to find simple yet effective, pre-configured solution packages that include all necessary hard- and software components, require little management effort and work out of the box. Fujitsu's PRIMEFLEX Cluster-in-a-box not only meets these demands, but has been around long enough to prove it helps IT departments to keep key applications and data permanently available – or, as the marketing guys would say: "to ensure continuous operation and high availability."

Astrid Hirschmüller

 

About the Author:

Astrid Hirschmüller

Product Manager for Cloud and Software-Defined Storage Solutions

SHARE

Comments on this article

No comments yet.

Please Login to leave a comment.