EC2 for In-Memory Computing - The High Memory Cluster Eight Extra Large Instance

Our new High Memory Cluster Eight Extra Large (cr1.8xlarge) instance type is designed to host applications that have a voracious need for compute power, memory, and network bandwidth such as in-memory databases, graph databases, and memory intensive HPC.

Here are the specs:

• Two Intel E5-2670 processors running at 2.6 GHz with Intel Turbo Boost and NUMA support.
• 244 GiB of RAM.
• Two 120 GB SSD for instance storage.
• 10 Gigabit networking with support for Cluster Placement Groups.
• HVM virtualization only.
• Support for EBS-backed AMIs only.

This is a real workhorse instance, with a total of 88 ECU (EC2 Compute Units). You can use it to run applications that are hungry for lots of memory and that can take advantage of 32 Hyperthreaded cores (16 per processor). We expect this instance type to be a great fit for in-memory analytics systems like SAP HANA and memory-hungry scientific problems such as genome assembly.

The Turbo Boost feature is very interesting. When the operating system requests the maximum possible processing power, the CPU increases the clock frequency while monitoring the number of active cores, the total power consumption and the processor temperature. The processor runs as fast as possible while staying within its documented temperature envelope.

NUMA (Non-Uniform Memory Access) speeds access to main memory by optimizing for workloads where the majority of requests for a particular block of memory come from one of the two processors. By enabling processor affinity (asking the scheduler to tie a particular thread to one of the processors) and taking care to manage memory allocation according to prescribed rules, substantial performance gains are possible. See this Intel article for more information on the use of NUMA.

Pricing starts at $3.50 per hour for Linux instances and $3.831 for Windows instances, both in US East (Northern Virginia). One year and three year Reserved Instances and Spot Instances are also available.

These instances are available in the US East (Northern Virginia) Region. We plan to make them available in other AWS Regions in the future.

-- Jeff;

Comments

[Terry A. Davis]

>MemRep;
>MemBIOSRep;

[Adrian Marius Popa]

One way to scale Firebird Database is to put it on memory virtual disk or ramfs

On linux i can mount a partition in memory (Install ubuntu/debian)

sudo mkdir /mnt/ram
sudo mount -t ramfs -o size=200G ramfs /mnt/ram
mount to show you the partitions mounted

and then move your database into the ram partition

cp -rp /var/lib/firebird/2.5/data /mnt/ram

That large instance could help you with the Firebird cache settings also the extra SSD could do wonders

Inspired by the stack overflow big fat server architecture

http://highscalability.com/blog/2009/8/5/stack-overflow-architecture.html