The true value of a successful innovation is not always immediately evident. But then it spawns more innovation, and its performance grows exponentially. That’s as true in cloud computing as it was a century ago in auto making. When Ransom Olds and Henry Ford began building cars on assembly lines, the early models were good. But the succeeding models quickly got better — by orders of magnitude. They were far faster, more reliable, and more affordable.
We’re climbing a similar innovation curve at Rackspace where, more than a year ago, we launched the world’s first public OpenStack cloud. That was, in a sense, a new kind of open, efficient factory for producing cloud servers. And today we’re unveiling our most powerful and elegant model yet — Performance Cloud Servers.
These servers have been engineered to deliver exceptional levels of application performance, with greater speed, throughput, reliability and scale. They are built with 100 percent data center-grade solid-state disks (SSDs). They incorporate powerful Intel® Xeon® E5 processors, as much as 120 GB of memory, and 40 gigabits per second of highly available network throughput to every host. They create a powerful hosting platform for a wide variety of workloads, ranging from basic web servers to large scale NoSQL data stores such as MongoDB and Cassandra.
In our benchmark tests, Performance Cloud Servers significantly outperformed comparatively sized Standard Cloud Servers. We saw as much as *:
In short, the workloads that run well on Cloud Servers today will run even better on Performance Cloud Servers. But that’s not all.
Performance Cloud Servers also offer larger sizes of compute and more total performance, compared to Standard Cloud Servers, including **:
This enables you to run workloads in the cloud, with high performance and efficiency, that could not run well — or in some cases, at all — on Standard Cloud Servers.
The performance gains enabled by this new product extend beyond the servers themselves. The Performance Cloud Servers’ high throughput network has been specifically designed to boost Cloud Block Storage performance, delivering as much as 1.5X more disk I/O for Standard volumes and 2.5X more disk I/O for SSD volumes ***. More network throughput also means improved performance on Cloud Databases, Cloud Load Balancers, and Cloud Files — and on dedicated infrastructure, linked to your Performance Cloud Servers via RackConnect.
These new cloud servers were designed for high reliability as well as performance. They run hardware RAID 10 and ECC memory, and include fully redundant network and power to every host.
We carefully engineered Performance Cloud Servers to maximize the performance you get at any price level. This allows us to offer the new servers for as little as two-third of the price of comparatively sized Standard Cloud Servers. That means you get more performance and reliability for less money.
As Performance Cloud Servers roll out, they will replace Standard Cloud Servers as our default offering. Performance Cloud Servers are available now in our Northern Virginia region. They will come online in our Dallas, Chicago, and London regions later in November. Our Sydney and Hong Kong regions will follow in the first half of 2014.
If you liked Cloud Servers, you’re going to love Performance Cloud Servers. If you’ve never tried Cloud Servers, now is the time. We are leveraging the power of OpenStack, the world’ s leading open source cloud operating system, to deliver extreme server performance on demand. And we’re doing so in a hybrid cloud environment that offers you the best-fit infrastructure for each of your workloads, whether it runs best on public cloud, private cloud, dedicated bare metal, or a combination of these platforms. We wrap everything up with fanatical expertise to help you whether you want to manage things yourself and just need a hand with our APIs, or you want to leverage our engineers to assist you in architecting, building, and running your infrastructure.
Welcome to the performance cloud!
* Benchmarks compare 1GB to 30GB Standard vs. Performance Cloud Servers. Linux kernel compile time was used to measure CPU performance, fio to measure disk I/O, iperf to measure network bandwidth, unixbench to measure overall performance, and YCSB to measure MongoDB performance. Multiple tests were run for each benchmark and the average score taken.
** Benchmarks compare a 30GB Standard Cloud Server to a 120GB Performance Cloud Server, the largest available in each class. Linux kernel compile time was used to measure CPU performance, fio to measure disk I/O, iperf to measure network bandwidth and unixbench to measure overall performance. Multiple tests were run for each benchmark and the average score taken.
*** Benchmarks use fio to compare Cloud Block Storage Standard and SSD volume performance on a 15GB Standard Cloud Server and 15GB Performance Cloud Server. Multiple tests were run and the average score taken.