Re: influencing the order in which page and swap files are used
From: Phillip Helbig---remove CLOTHES to reply (helbig_at_astro.multiCLOTHESvax.de)
Date: 01/06/04
- Next message: Phillip Helbig---remove CLOTHES to reply: "Re: Any using reset_errors or zero device error counters for Alpha?"
- Previous message: Richard B. Gilbert: "Re: Bugcheck - PCAPVCM?"
- In reply to: Keith A. Lewis: "Re: influencing the order in which page and swap files are used"
- Next in thread: dittman_at_dittman.net: "Re: influencing the order in which page and swap files are used"
- Reply: dittman_at_dittman.net: "Re: influencing the order in which page and swap files are used"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 5 Jan 2004 23:35:38 +0000 (UTC)
> >Actually I'm not rich, and have bought very little of my hobbyist
> >equipment. I picked up a few EZ32-VW disks for free, which I think will
> >work in a BA356 box, so why not? These have 256 MB, which is more than
> >the physical memory in any of my machines. The advantages are a) that I
> >already have the solid-state disk and b) that I can use it with any
> >system, whereas if I buy more memory it is tied to that box.
>
> Sounds reasonable to me. But be aware you're using a tank as a tractor.
> Solid-state disks have to be non-volatile, which means they're more
> expensive than regular RAM of the same speed. (FLASH, maybe?) Also the
> SCSI bus will probably have higher latency than RAM.
Right---real memory would be better, but my point is that the SSD is
much faster than a conventional disk, so fast that, as you say, the SCSI
adapter will be the bottleneck.
For a couple of reasons, I plan on moving some conventional disks on the
SCSI bus which would have the SSDs to other machines in the cluster.
(All disks are MSCP served and mounted by all nodes in the cluster.)
Presumably this will cut down on traffic on that SCSI bus without
increasing the network traffic (since the disks are members of shadow
sets and thus the commands to tell the other host node to write to the
disk or whatever already take place now)).
- Next message: Phillip Helbig---remove CLOTHES to reply: "Re: Any using reset_errors or zero device error counters for Alpha?"
- Previous message: Richard B. Gilbert: "Re: Bugcheck - PCAPVCM?"
- In reply to: Keith A. Lewis: "Re: influencing the order in which page and swap files are used"
- Next in thread: dittman_at_dittman.net: "Re: influencing the order in which page and swap files are used"
- Reply: dittman_at_dittman.net: "Re: influencing the order in which page and swap files are used"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|