Software-defined computing for in-memory big data workloads

Test Drive TidalScale Software-Defined Server Technology »

The Need for Software-Defined Servers

Today’s businesses run on data. The sources of that data and the velocity of that data are resulting in annual growth rates of 50% or more. Businesses need to process that data and act on those insights more and more quickly to stay competitive. In-memory computing is key to gaining rapid insights from data. Unfortunately, data is outgrowing the capabilities of today’s IT infrastructure.

TidalScale Software-Defined Server technology revolutionizes how in-memory workloads are deployed by breaking through the physical limitations of traditional server hardware while eliminating the complexity and operational overhead of moving to scale-out solutions. The result is massively scalable, composable systems optimized for in-memory computing that are deployed in minutes, work with the software you use today, and utilize readily available, cost-effective server and networking hardware.

TS-Wave
The Economics of Software-Defined Servers »
7 Technology Trends Driving the Industry to Software-Defined »
Guide to Software-Defined Servers in the Data Center »
Video:  Technology Overview »
Server Room
Video:  Technical Overview »
Video:  The Machine Learning Layer »
Video:  The Hyperkernel »
IEEE Scaling the Compute to the Problem »

Software-Defined Server Architecture

TidalScale aggregates the compute, memory and I/O resources of multiple physical servers into a single Software-Defined Server. A single Linux operating system image is deployed and ‘sees’ the virtual Software-Defined Server as it would a traditional scale-up server. No modifications to the operating system are necessary.

TidalScale brings composability to compute resources – Software-Defined Server technology utilizes a distributed, bare-metal hypervisor approach to virtualization. Servers are connected with commodity Ethernet networks and can be deployed, torn-down, and redeployed in minutes. Software-Defined Servers can deliver memory capacities from 10’s of Gigabytes to 10’s of Terabytes.

The TidalScale distributed hypervisor - the HyperKernel - is responsible for virtualizing and mobilizing all resources within the system. Resources continuously migrate within the system to maintain peak workload performance. Real time machine learning algorithms ensure optimal locality between memory pages, virtual CPUs, and system I/O.

Software-Defined Servers are deployed and operated just as a traditional servers. There are no changes to applications. They are container-friendly. Data does not need to be sharded. Software-Defined Servers integrate into high-availability (HA) frameworks just as scale-up servers would.

Deploying Software-Defined Servers: The WaveRunner® Console

Software-Defined Servers can be configured and deployed in minutes using TidalScale’s WaveRunner console, a GUI-based administration tool. The user simply identifies memory, compute, and storage desired and the WaveRunner tool draws Video:  WaveRunner  Demonstration » from the available pool of servers under administration and boots the Software-Defined Server. Pre-defined templates can be stored for different combinations of operating systems, processor core counts, memory, and storage. A Software-Defined Server can be up and running in as little as a few clicks and 5 minutes.

waverunner-on-laptop