back to article HPE adds power-fail-protected NVDIMM tech to servers

HPE's ProLiant Gen9 servers now support Persistent Memory, non-volatile memory, enabling customers to expand in-memory app usage and get faster analytics and database processing by avoiding disk and SSD IO. HPE says it will provide up to two-times-faster SQL Server database logging performance and up to four-times-faster SQL …

  1. Jon Massey
    WTF?

    96-watt-what?

    That's a decent amount of power but for how long?

    1. Anonymous Coward
      Anonymous Coward

      Re: 96-watt-what?

      Power is not required for long, just enough for the on board controller to copy the contents of the DRAM into the NAND that is on the NVDIMM, thus the persistence is not dependent on power availability beyond the very small time it takes to copy.

  2. Anonymous Coward
    Anonymous Coward

    huh?

    Is this so that a server can loose power and once its back on things start up right where things left off?

    What about data stuck in the CPU or in-between the CPU and IO bus, or one of the many other places that data can get temporarily cached as it moves thru a system board?

    At 8GB per DIMM and 128GB max per server, its certainly not a storage technology.

    Not bashing it, just not understanding its use case.

    1. DeepStorage

      Re: huh?

      It's the write cache in a storage system. If you're using these servers to build a scale-out storage system, hyper-converged or not, you can receive new data, write it to the NVDIMM, replicate to the NVDIMM in a second server and ACK. There are APIs you can use to ensure the data's been flushed from CPU cache to the NVDIM, and is therefore safe from power loss. Total time 250us or so.

      If you only have standard DIMMs you can't ACK until the data is someplace safe like a flash drive. Writing data to an SSD, even an NVMe SSD, will push overall latency up to 1-2MS.

  3. liquidh20

    For that price I could purchase a two fully redundant UPS systems, and two systems to run a synchronous cluster and still be cheaper than using this. This needs at least a 50% price drop, or maybe a single battery for multiple modules to reduce cost.

    1. luis river

      Hey, uncle... read this "HPE says it will provide up to two-times-faster SQL Server database logging performance and up to four-times-faster SQL cluster replications in block store mode. There is an up-to 63 per cent improvement in exchange speed, again on block mode."

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like