X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=UnderstandingLustre.xml;h=e8c2137849de4c68b044fe0f8b906cc6aef12615;hb=HEAD;hp=77690178b81b5dab0a1c1c1562c9dbeb9cd83e49;hpb=9ef599bdb59db2f59475f9b49177c3adc953e808;p=doc%2Fmanual.git
diff --git a/UnderstandingLustre.xml b/UnderstandingLustre.xml
index 7769017..e8c2137 100644
--- a/UnderstandingLustre.xml
+++ b/UnderstandingLustre.xml
@@ -133,18 +133,17 @@
Aggregate:
- 10 TB/sec I/O
+ 50 TB/sec I/O, 50M IOPSSingle client:
- 4.5 GB/sec I/O (FDR IB, OPA1),
- 1000 metadata ops/sec
+ 15 GB/sec I/O (HDR IB), 50000 IOPSAggregate:
- 2.5 TB/sec I/O
+ 10 TB/sec I/O, 10M IOPS
@@ -161,26 +160,26 @@
Single OST:
- 300M objects, 256TiB per OST (ldiskfs)
- 500M objects, 256TiB per OST (ZFS)
+ 500M objects, 1024TiB per OSTOSS count:
- 1000 OSSs, with up to 4000 OSTs
+ 1000 OSSs, 4000 OSTsSingle OSS:
- 32x 8TiB OSTs per OSS (ldiskfs),
- 8x 32TiB OSTs per OSS (ldiskfs)
- 1x 72TiB OST per OSS (ZFS)
+ 4 OSTs per OSS
+
+ Single OST:
+
+ 1024TiB OSTsOSS count:
- 450 OSSs with 1000 4TiB OSTs
- 192 OSSs with 1344 8TiB OSTs
- 768 OSSs with 768 72TiB OSTs
+ 450 OSSs with 900 750TiB HDD OSTs + 450 25TiB NVMe OSTs
+ 1024 OSSs with 1024 72TiB OSTs
@@ -193,21 +192,21 @@
Single OSS:
- 15 GB/sec
+ 15 GB/sec, 1.5M IOPSAggregate:
- 10 TB/sec
+ 50 TB/sec, 50M IOPSSingle OSS:
- 10 GB/sec
+ 10 GB/sec, 1.5M IOPSAggregate:
- 2.5 TB/sec
+ 20 TB/sec, 20M IOPS
@@ -224,22 +223,22 @@
Single MDT:
- 4 billion files, 8TiB per MDT (ldiskfs)
+ 4 billion files, 16TiB per MDT (ldiskfs)64 billion files, 64TiB per MDT (ZFS)MDS count:
- 256 MDSs, with up to 256 MDTs
+ 256 MDSs, up to 256 MDTsSingle MDS:
- 3 billion files
+ 4 billion filesMDS count:
- 7 MDS with 7 2TiB MDTs in production
+ 40 MDS with 40 4TiB MDTs in production256 MDS with 256 64GiB MDTs in testing
@@ -250,12 +249,12 @@
- 50000/s create operations,
- 200000/s metadata stat operations
+ 1M/s create operations
+ 2M/s stat operations
- 15000/s create operations,
- 50000/s metadata stat operations
+ 100k/s create operations,
+ 200k/s metadata stat operations
@@ -283,7 +282,7 @@
Aggregate:
- 55 PiB space, 8 billion files
+ 700 PiB space, 25 billion files
@@ -715,7 +714,7 @@
Verifies the linkEA entry for each inode and regenerates it if
invalid or missing. The linkEA
- stores of the file name and parent FID. It is stored as an extended
+ stores the file name and parent FID. It is stored as an extended
attribute in each inode. Thus, the linkEA can be used to
reconstruct the full path name of a file from only the FID.
@@ -732,7 +731,7 @@
striped across the objects using RAID 0,
and each object is stored on a different OST. (For more information about
how striping is implemented in a Lustre file system, see
- .
+ .