The IBM System x3650 M4 BD server is a dense 2U rack-mount system designed for Big Data (that’s why it’s called BD). Big Data is analytics on large unstructured data, and by big we mean really big. If you are serious about Big Data, then you might need as much as a Petabyte (1024 Terabytes) of local storage in a single rack. You’ll also need lots of processing power – lots of cores and lots of memory.
The x3650 M4 BD has what you need for Big Data:
High processing capacity
Large storage capacity
Energy efficiency to make your cluster as cost effective as possible
IBM System x3650 M4 BD
Twelve 3.5-inch drive bays at the front and two at the rear
Here are 5 things to know about this Big Data workhorse:
1. The x3650 M4 BD supports the latest Intel processors for high processing capacity with low power consumption
A server running Big Data applications needs to have high performance so that the analysis of the data completes as quickly as possible. A large number of cores is recommended, because it means greater processing concurrency – for example, you can configure more MapReduce slots per server with more cores.
The x3650 M4 BD server supports 2 processors, each with up to 12 cores, and memory up to 512 GB operating at speeds up to a blazing 2 DIMMs per channel (2DPC) at 1866 MHz. The Intel Xeon E5-2600 v2 processors offer outstanding performance while keeping power consumption as low as possible.
2. The x3650 M4 BD offers up to 56 TB of internal storage
Internal direct-attach storage is an essential part of any Big Data cluster offering and the x3650 M4 BD is designed with that requirement in mind. The server supports 14 drive bays, 12 at the front and 2 at the back. That sounds pretty good, but now consider that these are 3.5” drive bays, and each drive can be a 4 TB drive, which means a massive 56 TB of internal storage per 2U server – the largest capacity of any 2U 2 socket server in the industry. So with a 42U rack, you can have more than 1 Petabyte of raw data capacity!
One possible configuration would be to use the two rear drives as a RAID-1 pair for your operating system, and then the remaining 12 drives at the front can be configured simply as JBOD storage drives to maximize capacity. RAID is also offered for the local copies of the data being analyzed, should you need it. This gives you the flexibility, performance and capacity you need – data protection where you need it and maximum capacity where you need it.
The onboard controller supports 12 Gbps SAS, so as soon as 12 Gbps drives become available, the server will be ready to support them.
An additional storage choice is the IBM ServeRAID F5115 RAID controller with onboard flash storage, either 200GB or 800GB. This adapter is ideal for Big Data because it gives you high performance SSDs for hot data without consuming a drive bay. Connect the F5115 adapter to the 12 bays at the front of the server and the Intelligent caching algorithms of the adapter take care of managing the data.
The rear drive cage easily rotates up to give you access to the adapter slots.
Processors and memory are located under an air baffle.
3. The x3650 M4 BD is Energy Star 2.0 compliant
For Big Data clusters, energy efficiency is critical. Energy Star 2.0 is the new, more stringent, energy performance specifications defined by the Environmental Protection Agency (EPA) in the United State. The x3650 M4 BD met this guideline with the use of 80 PLUS Platinum power supplies, efficient components, and a carefully designed cooling subsystem. Passing this test means the server uses less energy, is less expensive to operate, and causes fewer greenhouse gas emissions than other non-compliant servers. Lower operating expense and better for the environment!
4. The x3650 M4 BD can operate in data centers up to 40°C
Having your servers use less power is one thing, but another part of your expense bottom line is the cost of operating your data center. The x3650 M4 BD is designed to operate in a warm data center, up to 40°C (104°F). This means you don’t have to chill your data center or server room as much, which means fewer CRAC units (lower CapEx) and smaller electricity bills (lower OpEx).
5. The x3650 M4 BD is the building block for Hadoop Based Big Data Software
Hadoop software from Independent Software Vendors and IBM, such as IBM InfoSphere BigInsights, lets you easily create Hadoop-based Big Data applications. This advanced software provides comprehensive data analytics capabilities, including administrative, discovery, development, provisioning, and security features. IBM InfosSphere Big Insights, as well as third party Hadoop software, runs on the x3650 M4 BD. For more information on IBM BigInsights, see the IBM InfoSphere BigInsights product page.
For more information on the IBM System x3650 M4 BD, see the IBM Redbooks Product Guide.
If you have 10 minutes, watch my video walk-through where I take you through the components of the server.
For an analyst’s view on the x3650 M4 BD plus other Big Data information, see the x3650 M4 BD home page.
David Watts is an IBM Redbooks Project Leader. He writes books and papers on many areas related to IBM Flex System, IBM BladeCenter and IBM System x. Follow David on Twitter at @DavidAtRedbooks.