enable small and large file optimization (#11260)

- for large objects we found that 1MiB block for
  r/w respectively.
- for small objects we found that 128KiB block for
  r/w respectively.
This commit is contained in:
Harshavardhana 2021-01-12 10:20:39 -08:00 committed by GitHub
parent e2579b1f5a
commit 1a5775e2e8
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 50 additions and 35 deletions

View file

@ -290,7 +290,7 @@ func TestGetObjectNoQuorum(t *testing.T) {
bucket := "bucket"
object := "object"
opts := ObjectOptions{}
buf := make([]byte, 33<<10)
buf := make([]byte, 129<<10)
if _, err = io.ReadFull(crand.Reader, buf); err != nil {
t.Fatal(err)
}

View file

@ -57,7 +57,7 @@ func (t *apiConfig) init(cfg api.Config, setDriveCount int) {
// max requests per node is calculated as
// total_ram / ram_per_request
// ram_per_request is 4MiB * setDriveCount + 2 * 10MiB (default erasure block size)
apiRequestsMaxPerNode = int(stats.TotalRAM / uint64(setDriveCount*(writeBlockSize+readBlockSize)+blockSizeV1*2))
apiRequestsMaxPerNode = int(stats.TotalRAM / uint64(setDriveCount*(blockSizeLarge+blockSizeSmall)+blockSizeV1*2))
} else {
apiRequestsMaxPerNode = cfg.RequestsMax
if len(globalEndpoints.Hostnames()) > 0 {

View file

@ -55,8 +55,8 @@ import (
const (
nullVersionID = "null"
diskMinTotalSpace = 900 * humanize.MiByte // Min 900MiB total space.
writeBlockSize = 4 * humanize.MiByte // Default write block size 4MiB.
readBlockSize = 2 * humanize.MiByte // Default read block size 2MiB.
blockSizeLarge = 1 * humanize.MiByte // Default r/w block size for larger objects.
blockSizeSmall = 128 * humanize.KiByte // Default r/w block size for smaller objects.
// On regular files bigger than this;
readAheadSize = 16 << 20
@ -65,9 +65,10 @@ const (
// Size of each buffer.
readAheadBufSize = 1 << 20
// Small file threshold below which data accompanies metadata
// from storage layer.
smallFileThreshold = 32 * humanize.KiByte
// Small file threshold below which data accompanies metadata from storage layer.
smallFileThreshold = 128 * humanize.KiByte // Optimized for NVMe/SSDs
// For hardrives it is possible to set this to a lower value to avoid any
// spike in latency. But currently we are simply keeping it optimal for SSDs.
// XL metadata file carries per object metadata.
xlStorageFormatFile = "xl.meta"
@ -97,8 +98,8 @@ type xlStorage struct {
globalSync bool
wpool sync.Pool
rpool sync.Pool
poolLarge sync.Pool
poolSmall sync.Pool
rootDisk bool
@ -261,15 +262,15 @@ func newXLStorage(ep Endpoint) (*xlStorage, error) {
p := &xlStorage{
diskPath: path,
endpoint: ep,
wpool: sync.Pool{
poolLarge: sync.Pool{
New: func() interface{} {
b := disk.AlignedBlock(writeBlockSize)
b := disk.AlignedBlock(blockSizeLarge)
return &b
},
},
rpool: sync.Pool{
poolSmall: sync.Pool{
New: func() interface{} {
b := disk.AlignedBlock(readBlockSize)
b := disk.AlignedBlock(blockSizeSmall)
return &b
},
},
@ -1180,7 +1181,7 @@ func (s *xlStorage) readAllData(volumeDir, filePath string) (buf []byte, err err
}
atomic.AddInt32(&s.activeIOCount, 1)
rd := &odirectReader{f, nil, nil, true, s, nil}
rd := &odirectReader{f, nil, nil, true, true, s, nil}
defer rd.Close()
return ioutil.ReadAll(rd)
@ -1410,6 +1411,7 @@ type odirectReader struct {
buf []byte
bufp *[]byte
freshRead bool
smallFile bool
s *xlStorage
err error
}
@ -1420,7 +1422,11 @@ func (o *odirectReader) Read(buf []byte) (n int, err error) {
return 0, o.err
}
if o.buf == nil {
o.bufp = o.s.rpool.Get().(*[]byte)
if o.smallFile {
o.bufp = o.s.poolSmall.Get().(*[]byte)
} else {
o.bufp = o.s.poolLarge.Get().(*[]byte)
}
}
if o.freshRead {
o.buf = *o.bufp
@ -1449,8 +1455,14 @@ func (o *odirectReader) Read(buf []byte) (n int, err error) {
// Close - Release the buffer and close the file.
func (o *odirectReader) Close() error {
o.s.rpool.Put(o.bufp)
atomic.AddInt32(&o.s.activeIOCount, -1)
if o.smallFile {
o.s.poolSmall.Put(o.bufp)
} else {
o.s.poolLarge.Put(o.bufp)
}
defer func() {
atomic.AddInt32(&o.s.activeIOCount, -1)
}()
return o.f.Close()
}
@ -1472,6 +1484,7 @@ func (s *xlStorage) ReadFileStream(ctx context.Context, volume, path string, off
}
var file *os.File
// O_DIRECT only supported if offset is zero
if offset == 0 && globalStorageClass.GetDMA() == storageclass.DMAReadWrite {
file, err = disk.OpenFileDirectIO(filePath, os.O_RDONLY, 0666)
} else {
@ -1512,7 +1525,10 @@ func (s *xlStorage) ReadFileStream(ctx context.Context, volume, path string, off
atomic.AddInt32(&s.activeIOCount, 1)
if offset == 0 && globalStorageClass.GetDMA() == storageclass.DMAReadWrite {
return &odirectReader{file, nil, nil, true, s, nil}, nil
if length <= smallFileThreshold {
return &odirectReader{file, nil, nil, true, true, s, nil}, nil
}
return &odirectReader{file, nil, nil, true, false, s, nil}, nil
}
if offset > 0 {
@ -1525,7 +1541,9 @@ func (s *xlStorage) ReadFileStream(ctx context.Context, volume, path string, off
io.Reader
io.Closer
}{Reader: io.LimitReader(file, length), Closer: closeWrapper(func() error {
atomic.AddInt32(&s.activeIOCount, -1)
defer func() {
atomic.AddInt32(&s.activeIOCount, -1)
}()
return file.Close()
})}
@ -1646,8 +1664,14 @@ func (s *xlStorage) CreateFile(ctx context.Context, volume, path string, fileSiz
w.Close()
}()
bufp := s.wpool.Get().(*[]byte)
defer s.wpool.Put(bufp)
var bufp *[]byte
if fileSize <= smallFileThreshold {
bufp = s.poolSmall.Get().(*[]byte)
defer s.poolSmall.Put(bufp)
} else {
bufp = s.poolLarge.Get().(*[]byte)
defer s.poolLarge.Put(bufp)
}
written, err := xioutil.CopyAligned(w, r, *bufp, fileSize)
if err != nil {

4
go.mod
View file

@ -66,7 +66,6 @@ require (
github.com/pierrec/lz4 v2.5.2+incompatible
github.com/pkg/errors v0.9.1
github.com/prometheus/client_golang v1.8.0
github.com/quasilyte/go-ruleguard v0.2.1 // indirect
github.com/rjeczalik/notify v0.9.2
github.com/rs/cors v1.7.0
github.com/secure-io/sio-go v0.3.0
@ -75,8 +74,7 @@ require (
github.com/streadway/amqp v1.0.0
github.com/tidwall/gjson v1.3.5
github.com/tidwall/sjson v1.0.4
github.com/tinylib/msgp v1.1.3
github.com/ttacon/chalk v0.0.0-20160626202418-22c06c80ed31 // indirect
github.com/tinylib/msgp v1.1.5
github.com/valyala/tcplisten v0.0.0-20161114210144-ceec8f93295a
github.com/willf/bitset v1.1.11 // indirect
github.com/willf/bloom v2.0.3+incompatible

15
go.sum
View file

@ -531,8 +531,6 @@ github.com/prometheus/procfs v0.0.8/go.mod h1:7Qr8sr6344vo1JqZ6HhLceV9o3AJ1Ff+Gx
github.com/prometheus/procfs v0.1.3/go.mod h1:lV6e/gmhEcM9IjHGsFOCxxuZ+z1YqCvr4OA4YeYWdaU=
github.com/prometheus/procfs v0.2.0 h1:wH4vA7pcjKuZzjF7lM8awk4fnuJO6idemZXoKnULUx4=
github.com/prometheus/procfs v0.2.0/go.mod h1:lV6e/gmhEcM9IjHGsFOCxxuZ+z1YqCvr4OA4YeYWdaU=
github.com/quasilyte/go-ruleguard v0.2.1 h1:56eRm0daAyny9UhJnmtJW/UyLZQusukBAB8oT8AHKHo=
github.com/quasilyte/go-ruleguard v0.2.1/go.mod h1:hN2rVc/uS4bQhQKTio2XaSJSafJwqBUWWwtssT3cQmc=
github.com/rcrowley/go-metrics v0.0.0-20181016184325-3113b8401b8a h1:9ZKAASQSHhDYGoxY8uLVpewe1GDZ2vu2Tr/vTdVAkFQ=
github.com/rcrowley/go-metrics v0.0.0-20181016184325-3113b8401b8a/go.mod h1:bCqnVzQkZxMG4s8nGwiZ5l3QUCyqpo9Y+/ZMZ9VjZe4=
github.com/rcrowley/go-metrics v0.0.0-20200313005456-10cdbea86bc0 h1:MkV+77GLUNo5oJ0jf870itWm3D0Sjh7+Za9gazKc5LQ=
@ -604,8 +602,8 @@ github.com/tidwall/pretty v1.0.0 h1:HsD+QiTn7sK6flMKIvNmpqz1qrpP3Ps6jOKIKMooyg4=
github.com/tidwall/pretty v1.0.0/go.mod h1:XNkn88O1ChpSDQmQeStsy+sBenx6DDtFZJxhVysOjyk=
github.com/tidwall/sjson v1.0.4 h1:UcdIRXff12Lpnu3OLtZvnc03g4vH2suXDXhBwBqmzYg=
github.com/tidwall/sjson v1.0.4/go.mod h1:bURseu1nuBkFpIES5cz6zBtjmYeOQmEESshn7VpF15Y=
github.com/tinylib/msgp v1.1.3 h1:3giwAkmtaEDLSV0MdO1lDLuPgklgPzmk8H9+So2BVfA=
github.com/tinylib/msgp v1.1.3/go.mod h1:+d+yLhGm8mzTaHzB+wgMYrodPfmZrzkirds8fDWklFE=
github.com/tinylib/msgp v1.1.5 h1:2gXmtWueD2HefZHQe1QOy9HVzmFrLOVvsXwXBQ0ayy0=
github.com/tinylib/msgp v1.1.5/go.mod h1:eQsjooMTnV42mHu917E26IogZ2930nFyBQdofk10Udg=
github.com/tmc/grpc-websocket-proxy v0.0.0-20170815181823-89b8d40f7ca8 h1:ndzgwNDnKIqyCvHTXaCqh9KlOWKvBry6nuXMJmonVsE=
github.com/tmc/grpc-websocket-proxy v0.0.0-20170815181823-89b8d40f7ca8/go.mod h1:ncp9v5uamzpCO7NfCPTXjqaC+bZgJeR0sMTm6dMHP7U=
github.com/ttacon/chalk v0.0.0-20160626202418-22c06c80ed31 h1:OXcKh35JaYsGMRzpvFkLv/MEyPuL49CThT1pZ8aSml4=
@ -626,7 +624,6 @@ github.com/xdg/stringprep v1.0.0/go.mod h1:Jhud4/sHMO4oL310DaZAKk9ZaJ08SJfe+sJh0
github.com/xiang90/probing v0.0.0-20190116061207-43a291ad63a2 h1:eY9dn8+vbi4tKz5Qo6v2eYzo7kUS51QINcR5jNpbZS8=
github.com/xiang90/probing v0.0.0-20190116061207-43a291ad63a2/go.mod h1:UETIi67q53MR2AWcXfiuqkDkRtnGDLqkBTpCHuJHxtU=
github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
github.com/yuin/goldmark v1.1.32/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
go.etcd.io/bbolt v1.3.3 h1:MUGmc65QhB3pIlaQ5bB4LwqSj6GIonVJXpZiaKNyaKk=
go.etcd.io/bbolt v1.3.3/go.mod h1:IbVyRI1SCnLcuJnV2u8VeU0CEYM7e686BmAb1XKL+uU=
@ -707,7 +704,6 @@ golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLL
golang.org/x/net v0.0.0-20200425230154-ff2c4b7c35a0/go.mod h1:qpuaurCH72eLCgpAm/N6yyVIVM9cpaDIP3A8BGJEC5A=
golang.org/x/net v0.0.0-20200625001655-4c5254603344/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
golang.org/x/net v0.0.0-20200707034311-ab3426394381/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
golang.org/x/net v0.0.0-20200822124328-c89045814202/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
golang.org/x/net v0.0.0-20200904194848-62affa334b73 h1:MXfv8rhZWmFeqX3GNZRsd6vOLoaCHjYEX3qkRo3YBUA=
golang.org/x/net v0.0.0-20200904194848-62affa334b73/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
@ -722,8 +718,7 @@ golang.org/x/sync v0.0.0-20181221193216-37e7f081c4d4/go.mod h1:RxMgew5VJxzue5/jJ
golang.org/x/sync v0.0.0-20190227155943-e225da77a7e6/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
golang.org/x/sync v0.0.0-20200625203802-6e8e738ad208 h1:qwRHBd0NqMbJxfbotnDhm2ByMI1Shq4Y6oRJo21SGJA=
golang.org/x/sync v0.0.0-20200625203802-6e8e738ad208/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9 h1:SQFwaSi55rU7vdNs9Yr0Z324VNlrF+0wMqRXT4St8ck=
golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
golang.org/x/sys v0.0.0-20180823144017-11551d06cbcc/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
golang.org/x/sys v0.0.0-20180830151530-49385e6e1522/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@ -795,9 +790,7 @@ golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtn
golang.org/x/tools v0.0.0-20191216052735-49a3e744a425/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28=
golang.org/x/tools v0.0.0-20200103221440-774c71fcf114/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28=
golang.org/x/tools v0.0.0-20200425043458-8463f397d07c/go.mod h1:EkVYQZoAsY45+roYkvgYkIh4xh/qjgUK9TdY2XT94GE=
golang.org/x/tools v0.0.0-20200812195022-5ae4c3c160a0/go.mod h1:njjCfa9FT2d7l9Bc6FUM5FLjQPp3cFF28FI3qnDFljA=
golang.org/x/tools v0.0.0-20200929223013-bf155c11ec6f h1:7+Nz9MyPqt2qMCTvNiRy1G0zYfkB7UCa+ayT6uVvbyI=
golang.org/x/tools v0.0.0-20200929223013-bf155c11ec6f/go.mod h1:z6u4i615ZeAfBE4XtMziQW1fSVJXACjjbWkB/mvPzlU=
golang.org/x/tools v0.0.0-20201022035929-9cf592e881e9/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
golang.org/x/tools v0.0.0-20210108195828-e2f9c7f1fc8e h1:Z2uDrs8MyXUWJbwGc4V+nGjV4Ygo+oubBbWSVQw21/I=
golang.org/x/tools v0.0.0-20210108195828-e2f9c7f1fc8e/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=