Re: Race Cache With Network experiment on Nightly
Em quarta-feira, 24 de maio de 2017 12:36:45 UTC-3, Valentin Gosu escreveu: > As part of the Quantum Network initiative we are working on a project > called "Race Cache With Network" (rcwn) [1]. > > This project changes the way the network cache works. When we detect that > disk IO may be slow, we send a network request in parallel, and we use the > first response that comes back. For users with slow spinning disks and a > low latency network, the result would be faster loads. > > This feature is currently preffed off - network.http.rcwn.enabled > In bug 1366224, which is about to land on m-c, we plan to enable it on > nightly for one or two days, to get some useful telemetry for our future > work. > > For any crashes or unexpected behaviour, please file bugs blocking 1307504. > > Thanks! > > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 Very useful information! Helped me in a lot. ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Re: Race Cache With Network experiment on Nightly
>I think you were looking at the docs for opt-in Shield studies Ah, right you are :) OK, I've filed a bug for the pref study: https://bugzilla.mozilla.org/show_bug.cgi?id=1367951 Thanks! Jason On Thu, May 25, 2017 at 2:27 PM, Matthew Grimeswrote: > I think you were looking at the docs for opt-in Shield studies > (experiments deployed as add-ons), not for pref flipping experiments. Due > to the nature of some of the opt-in studies we run they require a different > approval process. Pref flipping is available for all users, it is not > opt-in. The process currently requires one bug and an email to release > drivers. Feedback on the doc/process is always welcome! > > On Thu, May 25, 2017 at 1:40 PM, Jason Duell wrote: > >> I'm worried we're going from too little process here to too much (at >> least for this bug). Opening a meta-bug + 4 sub-bugs and doing a legal >> review, etc., is a lot of overhead to test some network plumbing that is >> not going to be especially noticeable to users. >> >> Also, we expect that this code will mostly benefit users with slow >> hardware (disk drives especially). We'll need to cast a very wide net to >> get nightly users that match that profile. The Shield docs say that >> "participation for Shield Studies is currently around 1-2% of randomly >> selected participants" (does that map to 1-2% of nightly users?), so I'm >> not sure we'd get enough coverage if we used Shield. >> >> Jason >> >> On Thu, May 25, 2017 at 11:30 AM, wrote: >> >>> Hey folks. I run the Shield team. Pref flipping experiments ARE >>> available on Nightly and will be available in all channels (including >>> Release) at some point in Firefox 54. >>> >>> Since the process is still relatively new, I've been hacking on some how >>> to docs: https://docs.google.com/document/d/16bpDZGCPKrOIgkkIo5mWKHPT >>> lYXOatyg_-CUi-3-e54/edit#heading=h.mzzhkdagng85 >>> >>> Feel free to give those a spin. Feedback on the docs/process is welcome. >>> >>> On Wednesday, May 24, 2017 at 6:14:55 PM UTC-7, Patrick McManus wrote: >>> > a howto for a pref experiment would be awesome.. >>> > >>> > On Wed, May 24, 2017 at 9:03 PM, Eric Rescorla wrote: >>> > >>> > > What's the state of pref experiments? I thought they were not yet >>> ready. >>> > > >>> > > -Ekr >>> > > >>> > > >>> > > On Thu, May 25, 2017 at 7:15 AM, Benjamin Smedberg < >>> benja...@smedbergs.us> >>> > > wrote: >>> > > >>> > > > Is there a particular reason this is landing directly to nightly >>> rather >>> > > > than using a pref experiment? A pref experiment is going to >>> provide much >>> > > > more reliable comparative data. In general we're pushing everyone >>> to use >>> > > > controlled experiments for nightly instead of landing experimental >>> work >>> > > > directly. >>> > > > >>> > > > --BDS >>> > > > >>> > > > On Wed, May 24, 2017 at 11:36 AM, Valentin Gosu < >>> valentin.g...@gmail.com >>> > > > >>> > > > wrote: >>> > > > >>> > > > > As part of the Quantum Network initiative we are working on a >>> project >>> > > > > called "Race Cache With Network" (rcwn) [1]. >>> > > > > >>> > > > > This project changes the way the network cache works. When we >>> detect >>> > > that >>> > > > > disk IO may be slow, we send a network request in parallel, and >>> we use >>> > > > the >>> > > > > first response that comes back. For users with slow spinning >>> disks and >>> > > a >>> > > > > low latency network, the result would be faster loads. >>> > > > > >>> > > > > This feature is currently preffed off - network.http.rcwn.enabled >>> > > > > In bug 1366224, which is about to land on m-c, we plan to enable >>> it on >>> > > > > nightly for one or two days, to get some useful telemetry for our >>> > > future >>> > > > > work. >>> > > > > >>> > > > > For any crashes or unexpected behaviour, please file bugs >>> blocking >>> > > > 1307504. >>> > > > > >>> > > > > Thanks! >>> > > > > >>> > > > > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn >>> > > > > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 >>> > > > > ___ >>> > > > > dev-platform mailing list >>> > > > > dev-platform@lists.mozilla.org >>> > > > > https://lists.mozilla.org/listinfo/dev-platform >>> > > > > >>> > > > ___ >>> > > > dev-platform mailing list >>> > > > dev-platform@lists.mozilla.org >>> > > > https://lists.mozilla.org/listinfo/dev-platform >>> > > > >>> > > ___ >>> > > dev-platform mailing list >>> > > dev-platform@lists.mozilla.org >>> > > https://lists.mozilla.org/listinfo/dev-platform >>> > > >>> > > >>> >>> ___ >>> dev-platform mailing list >>> dev-platform@lists.mozilla.org >>> https://lists.mozilla.org/listinfo/dev-platform >>> >> >> > ___ dev-platform mailing list
Re: Race Cache With Network experiment on Nightly
I think you were looking at the docs for opt-in Shield studies (experiments deployed as add-ons), not for pref flipping experiments. Due to the nature of some of the opt-in studies we run they require a different approval process. Pref flipping is available for all users, it is not opt-in. The process currently requires one bug and an email to release drivers. Feedback on the doc/process is always welcome! On Thu, May 25, 2017 at 1:40 PM, Jason Duellwrote: > I'm worried we're going from too little process here to too much (at least > for this bug). Opening a meta-bug + 4 sub-bugs and doing a legal review, > etc., is a lot of overhead to test some network plumbing that is not going > to be especially noticeable to users. > > Also, we expect that this code will mostly benefit users with slow > hardware (disk drives especially). We'll need to cast a very wide net to > get nightly users that match that profile. The Shield docs say that > "participation for Shield Studies is currently around 1-2% of randomly > selected participants" (does that map to 1-2% of nightly users?), so I'm > not sure we'd get enough coverage if we used Shield. > > Jason > > On Thu, May 25, 2017 at 11:30 AM, wrote: > >> Hey folks. I run the Shield team. Pref flipping experiments ARE available >> on Nightly and will be available in all channels (including Release) at >> some point in Firefox 54. >> >> Since the process is still relatively new, I've been hacking on some how >> to docs: https://docs.google.com/document/d/16bpDZGCPKrOIgkkIo5mWKHPT >> lYXOatyg_-CUi-3-e54/edit#heading=h.mzzhkdagng85 >> >> Feel free to give those a spin. Feedback on the docs/process is welcome. >> >> On Wednesday, May 24, 2017 at 6:14:55 PM UTC-7, Patrick McManus wrote: >> > a howto for a pref experiment would be awesome.. >> > >> > On Wed, May 24, 2017 at 9:03 PM, Eric Rescorla wrote: >> > >> > > What's the state of pref experiments? I thought they were not yet >> ready. >> > > >> > > -Ekr >> > > >> > > >> > > On Thu, May 25, 2017 at 7:15 AM, Benjamin Smedberg < >> benja...@smedbergs.us> >> > > wrote: >> > > >> > > > Is there a particular reason this is landing directly to nightly >> rather >> > > > than using a pref experiment? A pref experiment is going to provide >> much >> > > > more reliable comparative data. In general we're pushing everyone >> to use >> > > > controlled experiments for nightly instead of landing experimental >> work >> > > > directly. >> > > > >> > > > --BDS >> > > > >> > > > On Wed, May 24, 2017 at 11:36 AM, Valentin Gosu < >> valentin.g...@gmail.com >> > > > >> > > > wrote: >> > > > >> > > > > As part of the Quantum Network initiative we are working on a >> project >> > > > > called "Race Cache With Network" (rcwn) [1]. >> > > > > >> > > > > This project changes the way the network cache works. When we >> detect >> > > that >> > > > > disk IO may be slow, we send a network request in parallel, and >> we use >> > > > the >> > > > > first response that comes back. For users with slow spinning >> disks and >> > > a >> > > > > low latency network, the result would be faster loads. >> > > > > >> > > > > This feature is currently preffed off - network.http.rcwn.enabled >> > > > > In bug 1366224, which is about to land on m-c, we plan to enable >> it on >> > > > > nightly for one or two days, to get some useful telemetry for our >> > > future >> > > > > work. >> > > > > >> > > > > For any crashes or unexpected behaviour, please file bugs blocking >> > > > 1307504. >> > > > > >> > > > > Thanks! >> > > > > >> > > > > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn >> > > > > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 >> > > > > ___ >> > > > > dev-platform mailing list >> > > > > dev-platform@lists.mozilla.org >> > > > > https://lists.mozilla.org/listinfo/dev-platform >> > > > > >> > > > ___ >> > > > dev-platform mailing list >> > > > dev-platform@lists.mozilla.org >> > > > https://lists.mozilla.org/listinfo/dev-platform >> > > > >> > > ___ >> > > dev-platform mailing list >> > > dev-platform@lists.mozilla.org >> > > https://lists.mozilla.org/listinfo/dev-platform >> > > >> > > >> >> ___ >> dev-platform mailing list >> dev-platform@lists.mozilla.org >> https://lists.mozilla.org/listinfo/dev-platform >> > > ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Re: Race Cache With Network experiment on Nightly
I'm worried we're going from too little process here to too much (at least for this bug). Opening a meta-bug + 4 sub-bugs and doing a legal review, etc., is a lot of overhead to test some network plumbing that is not going to be especially noticeable to users. Also, we expect that this code will mostly benefit users with slow hardware (disk drives especially). We'll need to cast a very wide net to get nightly users that match that profile. The Shield docs say that "participation for Shield Studies is currently around 1-2% of randomly selected participants" (does that map to 1-2% of nightly users?), so I'm not sure we'd get enough coverage if we used Shield. Jason On Thu, May 25, 2017 at 11:30 AM,wrote: > Hey folks. I run the Shield team. Pref flipping experiments ARE available > on Nightly and will be available in all channels (including Release) at > some point in Firefox 54. > > Since the process is still relatively new, I've been hacking on some how > to docs: https://docs.google.com/document/d/16bpDZGCPKrOIgkkIo5mWKHPTlYXOa > tyg_-CUi-3-e54/edit#heading=h.mzzhkdagng85 > > Feel free to give those a spin. Feedback on the docs/process is welcome. > > On Wednesday, May 24, 2017 at 6:14:55 PM UTC-7, Patrick McManus wrote: > > a howto for a pref experiment would be awesome.. > > > > On Wed, May 24, 2017 at 9:03 PM, Eric Rescorla wrote: > > > > > What's the state of pref experiments? I thought they were not yet > ready. > > > > > > -Ekr > > > > > > > > > On Thu, May 25, 2017 at 7:15 AM, Benjamin Smedberg < > benja...@smedbergs.us> > > > wrote: > > > > > > > Is there a particular reason this is landing directly to nightly > rather > > > > than using a pref experiment? A pref experiment is going to provide > much > > > > more reliable comparative data. In general we're pushing everyone to > use > > > > controlled experiments for nightly instead of landing experimental > work > > > > directly. > > > > > > > > --BDS > > > > > > > > On Wed, May 24, 2017 at 11:36 AM, Valentin Gosu < > valentin.g...@gmail.com > > > > > > > > wrote: > > > > > > > > > As part of the Quantum Network initiative we are working on a > project > > > > > called "Race Cache With Network" (rcwn) [1]. > > > > > > > > > > This project changes the way the network cache works. When we > detect > > > that > > > > > disk IO may be slow, we send a network request in parallel, and we > use > > > > the > > > > > first response that comes back. For users with slow spinning disks > and > > > a > > > > > low latency network, the result would be faster loads. > > > > > > > > > > This feature is currently preffed off - network.http.rcwn.enabled > > > > > In bug 1366224, which is about to land on m-c, we plan to enable > it on > > > > > nightly for one or two days, to get some useful telemetry for our > > > future > > > > > work. > > > > > > > > > > For any crashes or unexpected behaviour, please file bugs blocking > > > > 1307504. > > > > > > > > > > Thanks! > > > > > > > > > > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn > > > > > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 > > > > > ___ > > > > > dev-platform mailing list > > > > > dev-platform@lists.mozilla.org > > > > > https://lists.mozilla.org/listinfo/dev-platform > > > > > > > > > ___ > > > > dev-platform mailing list > > > > dev-platform@lists.mozilla.org > > > > https://lists.mozilla.org/listinfo/dev-platform > > > > > > > ___ > > > dev-platform mailing list > > > dev-platform@lists.mozilla.org > > > https://lists.mozilla.org/listinfo/dev-platform > > > > > > > > ___ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform > ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Re: Race Cache With Network experiment on Nightly
Hey folks. I run the Shield team. Pref flipping experiments ARE available on Nightly and will be available in all channels (including Release) at some point in Firefox 54. Since the process is still relatively new, I've been hacking on some how to docs: https://docs.google.com/document/d/16bpDZGCPKrOIgkkIo5mWKHPTlYXOatyg_-CUi-3-e54/edit#heading=h.mzzhkdagng85 Feel free to give those a spin. Feedback on the docs/process is welcome. On Wednesday, May 24, 2017 at 6:14:55 PM UTC-7, Patrick McManus wrote: > a howto for a pref experiment would be awesome.. > > On Wed, May 24, 2017 at 9:03 PM, Eric Rescorlawrote: > > > What's the state of pref experiments? I thought they were not yet ready. > > > > -Ekr > > > > > > On Thu, May 25, 2017 at 7:15 AM, Benjamin Smedberg > > wrote: > > > > > Is there a particular reason this is landing directly to nightly rather > > > than using a pref experiment? A pref experiment is going to provide much > > > more reliable comparative data. In general we're pushing everyone to use > > > controlled experiments for nightly instead of landing experimental work > > > directly. > > > > > > --BDS > > > > > > On Wed, May 24, 2017 at 11:36 AM, Valentin Gosu > > > > > wrote: > > > > > > > As part of the Quantum Network initiative we are working on a project > > > > called "Race Cache With Network" (rcwn) [1]. > > > > > > > > This project changes the way the network cache works. When we detect > > that > > > > disk IO may be slow, we send a network request in parallel, and we use > > > the > > > > first response that comes back. For users with slow spinning disks and > > a > > > > low latency network, the result would be faster loads. > > > > > > > > This feature is currently preffed off - network.http.rcwn.enabled > > > > In bug 1366224, which is about to land on m-c, we plan to enable it on > > > > nightly for one or two days, to get some useful telemetry for our > > future > > > > work. > > > > > > > > For any crashes or unexpected behaviour, please file bugs blocking > > > 1307504. > > > > > > > > Thanks! > > > > > > > > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn > > > > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 > > > > ___ > > > > dev-platform mailing list > > > > dev-platform@lists.mozilla.org > > > > https://lists.mozilla.org/listinfo/dev-platform > > > > > > > ___ > > > dev-platform mailing list > > > dev-platform@lists.mozilla.org > > > https://lists.mozilla.org/listinfo/dev-platform > > > > > ___ > > dev-platform mailing list > > dev-platform@lists.mozilla.org > > https://lists.mozilla.org/listinfo/dev-platform > > > > ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Re: Race Cache With Network experiment on Nightly
What's the state of pref experiments? I thought they were not yet ready. -Ekr On Thu, May 25, 2017 at 7:15 AM, Benjamin Smedbergwrote: > Is there a particular reason this is landing directly to nightly rather > than using a pref experiment? A pref experiment is going to provide much > more reliable comparative data. In general we're pushing everyone to use > controlled experiments for nightly instead of landing experimental work > directly. > > --BDS > > On Wed, May 24, 2017 at 11:36 AM, Valentin Gosu > wrote: > > > As part of the Quantum Network initiative we are working on a project > > called "Race Cache With Network" (rcwn) [1]. > > > > This project changes the way the network cache works. When we detect that > > disk IO may be slow, we send a network request in parallel, and we use > the > > first response that comes back. For users with slow spinning disks and a > > low latency network, the result would be faster loads. > > > > This feature is currently preffed off - network.http.rcwn.enabled > > In bug 1366224, which is about to land on m-c, we plan to enable it on > > nightly for one or two days, to get some useful telemetry for our future > > work. > > > > For any crashes or unexpected behaviour, please file bugs blocking > 1307504. > > > > Thanks! > > > > [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn > > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 > > ___ > > dev-platform mailing list > > dev-platform@lists.mozilla.org > > https://lists.mozilla.org/listinfo/dev-platform > > > ___ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform > ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Re: Race Cache With Network experiment on Nightly
On 05/24/2017 03:36 PM, Valentin Gosu wrote: As part of the Quantum Network initiative we are working on a project called "Race Cache With Network" (rcwn) [1]. This project changes the way the network cache works. When we detect that disk IO may be slow, we send a network request in parallel, and we use the first response that comes back. For users with slow spinning disks and a low latency network, the result would be faster loads. Should we introduce a bias, in order to favor data coming from the alternate data, knowing that the alternate data stream might be bigger but reduce the upcoming work following the plain data? The concern here is that the alternate data might be slower to load, but faster to process. Do we have a way, from the nsICacheInfoChannel, to distinguish when data came from the network, as opposed to from the cache? The concern is that we want to avoid adding additional cost while try aggregate alternate data. Alternate cost which might slow down the execution speed after the network access. Overall this sounds like an awesome idea, if the stream is known to have no alternate data, or to compensate the alternate data savings. Maybe we could provide some estimates for the alternate data savings, based on the mime type? -- Nicolas B. Pierron ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Race Cache With Network experiment on Nightly
As part of the Quantum Network initiative we are working on a project called "Race Cache With Network" (rcwn) [1]. This project changes the way the network cache works. When we detect that disk IO may be slow, we send a network request in parallel, and we use the first response that comes back. For users with slow spinning disks and a low latency network, the result would be faster loads. This feature is currently preffed off - network.http.rcwn.enabled In bug 1366224, which is about to land on m-c, we plan to enable it on nightly for one or two days, to get some useful telemetry for our future work. For any crashes or unexpected behaviour, please file bugs blocking 1307504. Thanks! [1] https://bugzilla.mozilla.org/show_bug.cgi?id=rcwn [2] https://bugzilla.mozilla.org/show_bug.cgi?id=1366224 ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform