Serverless computing dynamically manages the compute and network resources that your application requires. Instead of you managing the fine details, DCP matches your workload with the right infrastructure.
DCP provides a unified API framework for your code to run anywhere, built to be intuitive for all developers. These can be called anywhere from a CL to a website.
DCP lets you run serverless applications on your own hardware, meaning fewer servers you have to manually configure.
Unlike other serverless solutions, DCP is entirely agnostic to the hardware owner. There is no requirement to migrate your data or build for a specific cloud platform.
Adjust your application in real time, so you can scale up or down without restarting work in progress. You can also get fine-grained runtime metrics to gauge and tune performance.
Hardware is always ready to execute your functions and begins working the instantly it receives instructions. For experienced serverless users, DCP solves the "cold start" problem.
Compute.for( ) simplifies your development workflow by automating how work is parallelized and distributed. By coordinating these tricky details, DCP is a point and click solution.
Not all data is created equal, and some is critical to protect. Therefore, DCP provides new tools for developers to keep data and algorithms safe given the unique nature of its distributed platform.
DCP expands your local capacity with secure clusters so you may never need to process in the cloud again. This keeps your data and algorithms within your LAN while benefiting from a serverless architecture.
For compute-intensive workloads with sensitive code or data, DCP lets you connect to trusted third party sites. Whether your data must remain in a jurisdiction or meet other security requirements, DCP will distribute it accordingly.
Protect data with secure SSH tunneling between servers when scaling beyond your LAN. Further, DCP implements cryptographic signatures on every message for tracking and nonrepudiation purposes.
By executing in a secure sandbox within the V8 JavaScript engine, code cannot interfere with the underlying device. This protects against XSS and CSRF vulnerabilities, and ensures workloads are completely isolated.
In addition to standard password and IAM methods, DCP enables Workers to request special identifiers from people accessing their hardware. This protects compute nodes from unauthorized users and applications.
DCP will implement native support for computing on encrypted data using the Microsoft SEAL library. Developers will be able to specify an appropriate level of encryption even though it may degrade compute performance.
Platforms have revolutionized the world, and many provide support for user made applications. DCP fully supports this, allowing users to build, share, and use 3rd party apps from finance to AI.
DCP apps can take on almost every frontend imaginable, from beautiful HTML5 pages to mobile apps. Your users can run advanced computations without a single line of code.
Build parameters into your apps that can be adapted for different experiments and datasets. DCP apps also allows others to back-check computations for effective collaboration.
Every app integrates with the DCP compute platform, whether on-premises or in the cloud. DCP also implements a deterministic math engine so results are identical across operating systems.
Apps are accessible on any device and require no download or installation. In the case of web apps, getting your audience set up is as easy as sharing a URL.
Apps can be free to use, require a license, or require payment based on compute consumption. Developers can seamlessly monetize their area of interest and make significant income.
DCP is built on JavaScript, with native support for NPM. With over 1m packages, the largest software registry in the world is ready to be used in your app.
Distributed computing must improve the performance of jobs while minimizing overhead. DCP solves the tough problems to make any collection of computers greater than the sum of its parts.
DCP's algorithms automatically optimize time- and space-sharing allocations for different jobs. This dynamically updates as jobs are added or change nature.
Easily estimate the size and compute requirements of your workload to accurately estimate its profile. DCP cross-validates results from different Workers for a high degree of confidence.
Build and manage your own personal compute cluster with spare servers and IoT devices. Access to these grids can be given on a permissioned basis, and they may be sub-divided into hyper-local networks.
Scavenge compute and memory from any machine with control over the time of day, type of application, and more. When resources are required locally, DCP instantly scales back.
The network is finally homogeneous with DCP. From 5G and SSH to Ethernet and open WiFi connections, DCP's overlay network accounts for differences in bandwidth and latency.
Developer needing control over how and where their applications are run. Therefore, DCP lets you simply manage how your code is executed under the same serverless framework.
Combine local Compute Groups on your LAN with ones in the cloud. DCP allows you to control remote nodes as easily as your local computer. You can even try different combinations with just a few clicks.
Process data close to the source. Anything from a smart home monitor to a connected vehicle can run your code, giving you almost infinite choice to build an edge-ready cluster.
Workloads adapt in real-time to the underlying compute nodes. This allows work to shift around when a node drops a connection, while livestreamed results mean a lost node does not mean lost time.
DCP is an ultra-lightweight framework with resting memory requirements as low as 0.1% of popular Container and VM services. Nodes can join and leave DCP clusters in an instant.
Offload computationally-intensive but stateless work onto DCP through your other applications. The API is built to be used in loosely coupled systems while adding almost no overhead.
Manage DCP with popular deployment options like Kubernetes, Containers, and VMs. DCP can scavenge unused compute from any hardware virtualization image, even when fully dedicated.
DCP is a unique tool for data parallel workloads of all shapes and sizes. Built by and for developers, its platform provides abundant compute from devices all around you.
The future of distributed computing is compute.for( ).
Built on open source technology, learn how DCP builds superclusters out of heterogeneous machines.
DCL intuitively incorporates some of the most advanced features of any compute platform. Learn how compute.for() changes everything.
