Re: [Openstack] Writes are faster than reads in Swift

2011-12-28 Thread Zhenhua (Gerald) Guo
Thanks for explaining it. I read the source code and confirmed that read operations incur more drop_cache calls than write operations. However, I am not sure whether the additional calls result in significant overhead. Need some tests to verify. Gerald On Tue, Dec 13, 2011 at 11:32 PM, Michael

Re: [Openstack] Writes are faster than reads in Swift

2011-12-28 Thread Zhenhua (Gerald) Guo
Thanks. Now, I understand the access pattern Swift is designed for. Do you know some real applications (or scenarios, use cases) that benefit from that design (except static image files)? Gerald On Wed, Dec 14, 2011 at 1:17 AM, Michael Barton mike-launch...@weirdlooking.com wrote: On Tue, Dec

Re: [Openstack] Writes are faster than reads in Swift

2011-12-15 Thread Jay Pipes
On Wed, Dec 14, 2011 at 5:38 PM, Ewan Mellor ewan.mel...@eu.citrix.com wrote: Or static disk image files! Only if you've got enough RAM on the storage worker node to cache the entire disk image.  Otherwise it's just going to get evicted straight away. The case where you've got so few,

Re: [Openstack] Writes are faster than reads in Swift

2011-12-15 Thread Ewan Mellor
are faster than reads in Swift On Wed, Dec 14, 2011 at 5:38 PM, Ewan Mellor ewan.mel...@eu.citrix.com wrote: Or static disk image files! Only if you've got enough RAM on the storage worker node to cache the entire disk image.  Otherwise it's just going to get evicted straight away. The case where

Re: [Openstack] Writes are faster than reads in Swift

2011-12-15 Thread Jay Pipes
On Thu, Dec 15, 2011 at 2:55 PM, Ewan Mellor ewan.mel...@eu.citrix.com wrote: Your simulation is of the one case that I said would work: when you have enough RAM to cache the entire image. When you _don't_ have enough RAM, then the images will just immediately be evicted, and the

Re: [Openstack] Writes are faster than reads in Swift

2011-12-14 Thread Jay Pipes
On Wed, Dec 14, 2011 at 1:17 AM, Michael Barton mike-launch...@weirdlooking.com wrote: On Tue, Dec 13, 2011 at 9:21 PM, Huang Zhiteng winsto...@gmail.com wrote: Can anyone explain why Swift doesn't want to utilize page cache _at all_? It's an artifact of the use case swift was built for -

Re: [Openstack] Writes are faster than reads in Swift

2011-12-14 Thread Ewan Mellor
: Re: [Openstack] Writes are faster than reads in Swift On Wed, Dec 14, 2011 at 1:17 AM, Michael Barton mike-launch...@weirdlooking.com wrote: On Tue, Dec 13, 2011 at 9:21 PM, Huang Zhiteng winsto...@gmail.com wrote: Can anyone explain why Swift doesn't want to utilize page cache _at all_

Re: [Openstack] Writes are faster than reads in Swift

2011-12-13 Thread Michael Barton
I can't explain it off the top of my head. I don't have a swift installation to play with at the moment, but it's conceivable that posix_fadvise is slower than we expect (drop_cache is called more frequently during reads than writes, iirc). That could be tested by making drop_cache a no-op in

Re: [Openstack] Writes are faster than reads in Swift

2011-12-13 Thread Huang Zhiteng
Can anyone explain why Swift doesn't want to utilize page cache _at all_? On Wed, Dec 14, 2011 at 12:32 PM, Michael Barton mike-launch...@weirdlooking.com wrote: I can't explain it off the top of my head. I don't have a swift installation to play with at the moment, but it's conceivable

Re: [Openstack] Writes are faster than reads in Swift

2011-12-13 Thread Michael Barton
On Tue, Dec 13, 2011 at 9:21 PM, Huang Zhiteng winsto...@gmail.com wrote: Can anyone explain why Swift doesn't want to utilize page cache _at all_? It's an artifact of the use case swift was built for - heavy on writes, and repeat reads (where a cache would help) are very rare. Having that