> "ag" == Andrew Gabriel writes:
ag> I can't speak for qmail which I've never used, but MTA's
ag> should sync data to disk before acknowledging receipt,
yeah, I saw a talk by one of the Postfix developers. They've taken
pains to limit the amount of sync'ing so it's only one or two c
Hi,
Am 12.10.2009 um 13:29 schrieb Richard Elling:
I've not implemented qmail, but it appears to be just an MTA.
These do store-and-forward, so it is unlikely that they need to
use sync calls. It will create a lot of files, but that is usually
done async.
Async I/O for mail servers is a big n
Richard Elling wrote:
On Oct 12, 2009, at 2:12 AM, tak ar wrote:
I'm not aware of email services using sync regularly.
In my experience
with large
email services, the response time of the disks used
for database and
indexes is
the critical factor (for > 600 messages/sec
delivered, caches don't
On Oct 12, 2009, at 2:12 AM, tak ar wrote:
I'm not aware of email services using sync regularly.
In my experience
with large
email services, the response time of the disks used
for database and
indexes is
the critical factor (for > 600 messages/sec
delivered, caches don't
matter :-)
Performance
> I'm not aware of email services using sync regularly.
> In my experience
> with large
> email services, the response time of the disks used
> for database and
> indexes is
> the critical factor (for > 600 messages/sec
> delivered, caches don't
> matter :-)
> Performance of the disks for the
> > Use the BBWC to maintain high IOPS when X25-E's
> write cache is disabled?
>
> It should certainly help. Note that in this case
> your relatively
> small battery-backed memory is accepting writes for
> both the X25-E and
> for the disk storage so the BBWC memory becomes 1/2
> as useful and
> It is currently difficult to remove slog devices so it is safer to add
> them if you determine they will help rather than reduce performance.
>
> Bob
Fixed in 125 bob:
http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6574286
--
This message posted from opensolaris.org
__
On Oct 10, 2009, at 4:11 PM, Bob Friesenhahn wrote:
On Sat, 10 Oct 2009, tak ar wrote:
I think the IOPS is important for mail server, so ZIL is useful.
The server has 48GB RAM and two(ZFS or hardware mirror) X25-E(32GB)
for ZIL(slog). I understand the ZIL needs half of RAM.
There is a dif
On Sat, 10 Oct 2009, tak ar wrote:
Use the BBWC to maintain high IOPS when X25-E's write cache is disabled?
It should certainly help. Note that in this case your relatively
small battery-backed memory is accepting writes for both the X25-E and
for the disk storage so the BBWC memory becomes
Hi! thanks for reply.
> The BBWC is much more useful than the write cache on
> the X25-E since
> the X25-E's write cache is volatile and therefore may
> cause harm to
> your data. According to reports I have seen, the
> X25-E write IOPS
> reduces by a factor of five when its write cache is
> d
On Fri, 9 Oct 2009, tak ar wrote:
Hi! I bought x4270 servers for (write heavy) mail server. And
waiting for delivery. That have two Intel SSD X25-E(for ZIL) and
HDDs. x4270 servers have hardware RAID card based on Adaptec's RAID
5805 adapter, which has 256MB BBWC.
SSD has write cache and RAI
On Fri, Oct 9 at 22:51, tak ar wrote:
When the answer is no, should I set disable SSD's write cache? I
think disabled write cache reduce the usable lifetime of
SSD. Because wear-leveling on SSD is not applied.
I don't think their wear leveling requires the write cache to be enabled.
--
Eric D
Hi! I bought x4270 servers for (write heavy) mail server. And waiting for
delivery. That have two Intel SSD X25-E(for ZIL) and HDDs. x4270 servers have
hardware RAID card based on Adaptec's RAID 5805 adapter, which has 256MB BBWC.
SSD has write cache and RAID card also has BBWC. When set write-
13 matches
Mail list logo
