Re: [Patch][WiP] Tweaked LRU for shared buffers

Hi Tomas, If you are on a benchmarking binge and feel like generating some trace files (as mentioned earlier), I'd be happy to help in regards to running them through simulations to show how different policies behave. We can add more types to match this patch / Postgres' GClock as desired, too. O

Re: [Patch][WiP] Tweaked LRU for shared buffers

On 2/17/19 2:14 PM, Едигарьев, Иван Григорьевич wrote: > Hi there. I was responsible for the benchmarks, and I would be glad to > make clear that part for you. > > On Sat, 16 Feb 2019 at 02:30, Tomas Vondra > wrote: >> Interesting. Where do these numbers (5/8 and 1/8) come from? > > The first n

Re: [Patch][WiP] Tweaked LRU for shared buffers

Hi there. I was responsible for the benchmarks, and I would be glad to make clear that part for you. On Sat, 16 Feb 2019 at 02:30, Tomas Vondra wrote: > Interesting. Where do these numbers (5/8 and 1/8) come from? The first number came from MySQL realization of LRU algorithm

Re: [Patch][WiP] Tweaked LRU for shared buffers

On 2/16/19 10:36 AM, Vladimir Sitnikov wrote: > Benjamin> A related and helpful patch would be to capture the access log and > Benjamin> provide anonymized traces. > > The traces can be captured via DTrace scripts, so no patch is required here. > Right. Or a BPF on reasonably new linux kernels.

Re: [Patch][WiP] Tweaked LRU for shared buffers

Benjamin> A related and helpful patch would be to capture the access log and Benjamin> provide anonymized traces. The traces can be captured via DTrace scripts, so no patch is required here. For instance: https://www.postgresql.org/message-id/CAB%3DJe-F_BhGfBu1sO1H7u_XMtvak%3DBQtuJFyv8cfjGBRp7Q_y

Re: [Patch][WiP] Tweaked LRU for shared buffers

Hi, I was not involved with Andrey and his team's work, which looks like a very promising first pass. I can try to clarify a few minor details. What is CAR? Did you mean ARC, perhaps? CAR is the Clock variants of ARC: CAR: Clock with Adaptive Replacement

Re: [Patch][WiP] Tweaked LRU for shared buffers

> > No, I "synchronized scans" are an optimization to reduce I/O when multiple > processes do sequential scan on the same table. Oh, very neat. Thanks! Interesting. I assume the trace is essentially a log of which blocks were > requested? Is there some trace format specification somewhere? > Ye

Re: [Patch][WiP] Tweaked LRU for shared buffers

On 2/16/19 1:48 AM, Benjamin Manes wrote: > Hi, > > I was not involved with Andrey and his team's work, which looks like a > very promising first pass. I can try to clarify a few minor details. > > What is CAR? Did you mean ARC, perhaps? > > > CAR is the Clock variants of ARC: CAR: Clock wi

Re: [Patch][WiP] Tweaked LRU for shared buffers

On 2/16/19 12:51 AM, Peter Geoghegan wrote: > On Fri, Feb 15, 2019 at 3:30 PM Tomas Vondra > wrote: >> That TPS chart looks a bit ... wild. How come the master jumps so much >> up and down? That's a bit suspicious, IMHO. > > Somebody should write a patch to make buffer eviction completely > ra

Re: [Patch][WiP] Tweaked LRU for shared buffers

On Fri, Feb 15, 2019 at 3:30 PM Tomas Vondra wrote: > That TPS chart looks a bit ... wild. How come the master jumps so much > up and down? That's a bit suspicious, IMHO. Somebody should write a patch to make buffer eviction completely random, without aiming to get it committed. That isn't as bad

Re: [Patch][WiP] Tweaked LRU for shared buffers

Hi, On 2/13/19 3:37 PM, Andrey Borodin wrote: > Hi, hackers! > > We have held education project at Sirius edu center (Sochi, Russia) > with mentors from Yandex. The group of 5 students was working on > improving the shared buffers eviction algorithm: Andrey Chausov, Yuriy > Skakovskiy, Ivan Edi

[Patch][WiP] Tweaked LRU for shared buffers

Hi, hackers! We have held education project at Sirius edu center (Sochi, Russia) with mentors from Yandex. The group of 5 students was working on improving the shared buffers eviction algorithm: Andrey Chausov, Yuriy Skakovskiy, Ivan Edigaryev, Arslan Gumerov, Daria Filipetskaya. I’ve been a me