PiCloud gives every scientist, developer, and engineer in the world access to clusters of machines that 10 years ago were available only to the largest companies and national laboratories.
The PiCloud Platform is designed for complex science, engineering and business applications that require immense computational power, large data storage, and high network bandwidth.
Our platform provides an easy-to-use, one-stop solution that can replace or supplement your existing server hardware, batch scheduler, orchestration software, and data storage. Our customers have reduced their infrastructure costs by over 50%, sped up processing by 1,000 fold, and saved thousands of man-hours in development time.
To use PiCloud, you integrate your applications with our API. The two solutions we offer differ in whether you want to use hardware managed by PiCloud (public), or your own hardware (private).
The PiCloud that you can use today from our website uses servers from Amazon Web Services. To use PiCloud, you won’t need any hardware or server-side installation; just sign up and install our client library on your machines. You will be able to scale to thousands of cores on-demand without owning a single server.
Our private cloud solution can be installed in your own datacenter running Linux or Windows. Our solution can work alongside other batch schedulers including Oracle Grid Engine (formerly Sun Grid Engine) and Platform LSF. For our private cloud solution, your scalability is bounded by the number of machines and cores you have.
The PiCloud Platform has 3 key components explained briefly below. For more information, see our Platform page.
Our compute service allows us to scale processing to thousands of heterogeneous nodes with sub-second overhead. This includes scheduling, scaling, and load balancing.
We include two data storage offerings, and the platform can inter-operate with most other data storage solutions you may already be using. Buckets give you a key->vlaue store with scalability to petabytes of data. Volumes give you the ability to use a distributed file system with terabytes of data, which to your legacy applications appear local to the machine for maximum backwards compatibility.
The customization engine lets you deploy any Linux-based software binaries or libraries without modification. This allows you to use any programming language and any software dependencies on PiCloud with the automation benefits of our platform.
The following is only applicable to the public cloud solution.
In-house clusters require committing to hundreds of thousands to millions of dollars in capital expenditures. With PiCloud, capital expenditures (CapEx) are converted to operational expenditures (OpEx). You only have to pay for as much as is used, greatly minimizing your investment risk. And in five years, you won’t be stuck with a datacenter full of aging hardware.
Have a big project that needs 2,000 cores instead of the 1,000 in your datacenter? Or, have two projects with similar deadlines that each need 1,000 cores? With your own datacenter, your options are slim. Time sharing and planning months in advance are a must. With PiCloud, you can provision thousands of additional cores on-demand, and only minutes in advance.
With your own cluster, you’ll have dedicated staff working to maintain hardware, replace failed components, and upgrade. On top of the hardware, staff will need to configure operating systems, standardize packages, apply updates, and provide software support.
PiCloud obviates all hardware management, and has facilities to standardize the software packages (Environments) users can use. When assistance is needed, users can call our phone support line.
We've focused on making PiCloud the easiest to use computing platform in existence. As you can see from our documentation, after installing our client, you can be running jobs with input/output data and full monitoring, in as few as three lines.
Batch schedulers require you to setup each machine with access to all the software and data needed for computation. Updating your software and data requires you to re-deploy to your machines. Using Environments, PiCloud is able to version, deploy, and cache resources to severs automatically.
Batch schedulers traditionally leave it up to your organization to setup and manage a data store that can be accessed by your cluster. Using a global NFS store is often the solution due to its ease of configuration. Using a global NFS store, however, has two drawbacks for HPC.
PiCloud offers two scalable storage solutions: Buckets, an object store, that can house petabytes of data; and Volumes, an NFS-like distributed file system. Both solutions are redundant for high availability, and parallel read throughput performance. Lastly, no administrator is needed to maintain these storage systems.
Scientific Python users will feel at home with PiCloud's deep integration into the language. In addition, for all users, PiCloud has over 500 packages installed by default so that moving scientific computing applications to the cloud is seamless.
Since our public cloud solution is open to everyone--not hidden behind an "enterprise portal"--our solutions have been tested by thousands of users worldwide, who have collectively processed over 100 million jobs.
Never worry again about underutilizing your fleet of servers. Because our public cloud solution is on-demand, and charges you by the millisecond, we’ve made it our responsibility to ensure efficient load balancing and high utilization.
See our platform page for the capabilities we offer.
Amazon EC2 provides no comparable solution to our private cloud solution, which can be installed in your datacenter. For a comparison of our public cloud solution, please see our platform page.
You can take advantage of remote support services via telephone, virtual tickets, and e-mail. For mission critical applications, you can subscribe to 24 x 7 x 365 day support with a 1-hour maximum reponse time.
If you're looking for more than support using PiCloud, we offer consulting services. We've worked with customers to achieve goals including:
For more information, please send us an inquiry.
Our public cloud prices for core hours, storage, and bandwidth are listed on the Pricing page. All billing is done on a pay-as-you-go basis. For customers with expected computation bills over $1,000/month, we offer tailored discounts ranging from 10% to 40% based on usage patterns and frequency. Customers can also subscribe to monthly and annual package plans for 10% and 35% discounts, respectively.
For on-premise installations, we charge a license fee on a per core per year basis. More specifically, we charge for cores being "actively managed," which means you can install PiCloud on all of your internal machines, making them all candidates for running workloads. The number of licenses you have determines the number of machines that will simultaneously run workloads.
We provide quotes for private cloud installations after a thorough review of your internal infrastructure to determine the complexity of deploying our software in your environment. The prices for our public cloud offering do not apply to our private cloud offering. If interested, please send us an inquiry.
Whether you’re interested in the public or private cloud solution, you should consider signing up for a free public cloud account. The account will provide the same interface as you would see for either solution, and allow you to test your workloads. See our documentation.
We're happy to speak with you over e-mail, Skype, or phone to understand your needs, and how we can help. Just fill out the form below, and we'll get in touch with you: