To meet the growing memory demand from AI, Big Data workloads, virtualization, and high-performance computing, today’s data centers require expensive RAM-filled servers. However, data centers never use this traditional memory 100% of the time. This unused memory is stranded inside servers — bonded to particular CPUs, unsharable, and isolated from underutilized resources elsewhere in the data center.
Kove® External Memory revolutionizes how developers, administrators, and data centers think about memory by breaking down the physical barriers of traditional server hardware and decoupling memory from the server. The result? A massively scalable, provisionable memory resource just as fast as local memory. Kove® External Memory is easy to configure, requires no application code changes, and utilizes cost-effective commodity servers.
Kove® External Memory makes it easy to quickly turn multiple physical servers into a single software-defined memory pool. Applications use the memory pool as they would traditional server memory, no code changes required. Just add more servers and the memory pool scales linearly, without performance degradation.
Kove’s Management Console makes it easy to administer the memory pool. Adding servers to the pool takes a few minutes and easy-to-configure rules control memory provisioning.
Applications using traditional memory or storage methods can use the memory pool without any code changes. For maximum performance, applications may also use Kove’s APIs which bypass the kernel altogether. Of course, virtualization platforms seamlessly integrate with the memory pool, enabling massive flexibility and scale.
Just install Kove® EMX software on a commodity server and use the Management Console to add the server to the memory pool. The server’s once stranded memory joins the pool and becomes available for use whenever and wherever across the data center.
Kove® External Memory delivers deterministically low latency and, in most cases, statistically matches CPU needs for local memory, allowing consistent CPU saturation. Because Kove® External Memory enables limitless memory sizing, any size computation can run completely in memory.
Data centers never use traditional memory 100% of the time, resulting in wide-spread memory stranding. Kove® External Memory reduces memory stranding by separating memory from the server. As a result, data centers need fewer servers with less memory, and upgrades to next-gen CPUs no longer require repurchasing the same memory. This reduces all associated operating costs, such as power, cooling, software, and even personnel. ROI in the 50-80% range is common.
Today’s data centers require expensive RAM-filled servers for “big memory” use cases but data centers rarely fully utilize this memory. Servers also bundle memory and CPU together, forcing the repurchase of the same amount of memory just to upgrade CPUs. With Kove® External Memory, individual servers need less memory, making frequent and cheaper server upgrades to next-gen CPUs possible. Server lifespans can even be extended by adding older servers to the memory pool.
With Kove® External Memory, memory capacity grows beyond the limits of local physical DIMM slots, allowing seamless access to memory up to CPU-addressable limits. For example, allocate 2.25 TiB of memory to any of 200 servers for nightly jobs, create a virtual machine with larger memory than the physical hypervisor, or allocate 40 TiB to a single server for an hour for burst ingest. When jobs complete, memory returns to the pool, available for allocation to new servers and jobs.
