Differences

This shows you the differences between two versions of the page.

--- pandoc:introduction-to-vsc:08_storage_infrastructure:storage_infrastructure [2018/01/31 11:10] – Pandoc Auto-commit pandoc
+++ pandoc:introduction-to-vsc:08_storage_infrastructure:storage_infrastructure [2020/10/20 09:13] (current) – Pandoc Auto-commit pandoc
@@ Line 1: / Line 1: @@
+====== Storage infrastructure ======
+  * Article written by Siegfried Reinwald (VSC Team) <html><br></html>(last update 2019-01-15 by sh).
+====== Storage hardware VSC-3 ======
+  * Storage on VSC-3
+    * 10 Servers for ''%%$HOME%%''
+    * 8 Servers for ''%%$GLOBAL%%''
+    * 16 Servers for ''%%$BINFS%%'' / ''%%$BINFL%%''
+    * ~ 800 spinning disks
+    * ~ 100 SSDs
+====== Storage targets ======
+  * Several Storage Targets on VSC-3
+    * ''%%$HOME%%''
+    * ''%%$TMPDIR%%''
+    * ''%%$SCRATCH%%''
+    * ''%%$GLOBAL%%''
+    * ''%%$BINFS%%''
+    * ''%%$BINFL%%''
+  * For different purposes
+    * Random I/O
+    * Small Files
+    * Huge Files / Streaming Data
+====== Storage performance ======
+{{.:vsc3_storage_performance.png}}
+====== The HOME Filesystem (VSC-3) ======
+  * Use for non I/O intensive jobs
+  * Basically NFS Exports over infiniband (no RDMA)
+  * Logical volumes of projects are distributed among the servers
+    * Each logical volume belongs to 1 NFS server
+  * Accessible with the ''%%$HOME%%'' environment variable
+    * /home/lv70XXX/username
+====== The GLOBAL and SCRATCH filesystem ======
+  * Use for I/O intensive jobs
+  * ~ 500 TB Space (default quota is 500GB/project)
+    * Can be increased on request (subject to availability)
+  * BeeGFS Filesystem
+  * Accessible via the ''%%$GLOBAL%%'' and ''%%$SCRATCH%%'' environment variables
+    * ''%%$GLOBAL%%'' … ///global/lv70XXX/username//
+    * ''%%$SCRATCH%%'' … ///scratch//
+  * Check quota
+<code>
+    beegfs-ctl --getquota --cfgFile=/etc/beegfs/global3.d/beegfs-client.conf --gid 70XXX
+</code>
+<code>
+VSC-3 > beegfs-ctl --getquota --cfgFile=/etc/beegfs/global3.d/beegfs-client.conf --gid 70824
+      user/group     ||           size          ||    chunk files
+     name     |  id  ||    used    |    hard    ||  used   |  hard
+--------------|------||------------|------------||---------|---------
+        p70824| 70824||      0 Byte|  500.00 GiB||        0|   100000
+</code>
+====== The BINFL filesystem ======
+  * Specifically designed for Bioinformatics applications
+  * Use for I/O intensive jobs
+  * ~ 1 PB Space (default quota is 10GB/project)
+    * Can be increased on request (subject to availability)
+  * BeeGFS Filesystem
+  * Accessible via ''%%$BINFL%%'' environment variable
+    * ''%%$BINFL%%'' … ///binfl/lv70XXX/username//
+  * Also available on VSC-4
+  * Check quota
+<code>
+    beegfs-ctl --getquota --cfgFile=/etc/beegfs/hdd_storage.d/beegfs-client.conf --gid 70XXX
+</code>
+<code>
+VSC-3 > beegfs-ctl --getquota --cfgFile=/etc/beegfs/hdd_storage.d/beegfs-client.conf --gid 70824
+      user/group     ||           size          ||    chunk files
+     name     |  id  ||    used    |    hard    ||  used   |  hard
+--------------|------||------------|------------||---------|---------
+        p70824| 70824||    5.93 MiB|   10.00 GiB||      574|  1000000
+</code>
+====== The BINFS filesystem ======
+  * Specifically designed for Bioinformatics applications
+  * Use for very I/O intensive jobs
+  * ~ 100 TB Space (default quota is 2GB/project)
+    * Can be increased on request (subject to availability)
+  * BeeGFS Filesystem
+  * Accessible via ''%%$BINFS%%'' environment variable
+    * ''%%$BINFS%%'' … ///binfs/lv70XXX/username//
+  * Also available on VSC-4
+  * Check quota
+<code>
+    beegfs-ctl --getquota --cfgFile=/etc/beegfs/nvme_storage.d/beegfs-client.conf --gid 70XXX
+</code>
+<code>
+VSC-3 > beegfs-ctl --getquota --cfgFile=/etc/beegfs/nvme_storage.d/beegfs-client.conf --gid 70824
+      user/group     ||           size          ||    chunk files
+     name     |  id  ||    used    |    hard    ||  used   |  hard
+--------------|------||------------|------------||---------|---------
+        p70824| 70824||      0 Byte|    2.00 GiB||        0|     2000
+</code>
+====== The TMP filesystem ======
+  * Use for
+    * Random I/O
+    * Many small files
+  * Size is up to 50% of main memory
+  * Data gets deleted after the job
+    * Write Results to ''%%$HOME%%'' or ''%%$GLOBAL%%''
+  * Disadvantages
+    * Space is consumed from main memory <html><!--* Alternatively the mmap() system call can be used
+  * Keep in mind, that mmap() uses lazy loading
+  * Very small files waste main memory (memory mapped files are aligned to page-size)--></html>
+  * Accessible with the ''%%$TMPDIR%%'' environment variable
+====== Storage hardware VSC-4 ======
+  * Storage on VSC-4
+    * 1 Server for ''%%$HOME%%''
+    * 6 Servers for ''%%$DATA%%''
+    * 720 spinning disks
+    * 16 NVMEs flash drives
+====== The HOME Filesystem (VSC-4) ======
+  * Use for software and job scripts
+  * Default quota: 100GB
+  * Accessible with the ''%%$HOME%%'' environment variable (VSC-4)
+    * /home/fs70XXX/username
+  * Also available on VSC-3
+    * /gpfs/home/fs70XXX/username
+  * Check quota
+<code>
+mmlsquota --block-size auto -j home_fs70XXX home
+</code>
+<code>
+VSC-4 > mmlsquota --block-size auto -j home_fs70824 home
+                         Block Limits                                    |     File Limits
+Filesystem type         blocks      quota      limit   in_doubt    grace |    files   quota    limit in_doubt
+home       FILESET       63.7M       100G       100G          0     none |     3822 1000000  1000000        0
+</code>
+====== The DATA Filesystem ======
+  * Use for all kind of I/O
+  * Default quota: 10TB
+    * Extansion can be requested
+  * Accessible with the ''%%$DATA%%'' environment variable (VSC-4)
+    * /data/fs70XXX/username
+  * Also available on VSC-3
+    * /gpfs/data/fs70XXX/username
+  * Check quota
+<code>
+mmlsquota --block-size auto -j data_fs70XXX data
+</code>
+<code>
+VSC-4 > mmlsquota --block-size auto -j data_fs70824 data
+                         Block Limits                                    |     File Limits
+Filesystem type         blocks      quota      limit   in_doubt    grace |    files   quota    limit in_doubt
+data       FILESET           0     9.766T     9.766T          0     none |       14 1000000  1000000        0
+</code>
+====== Backup policy ======
+  * Backup of user files is **solely the responsibility of each user**
+    * [[https://service.vsc.ac.at/slides/introduction-to-vsc/02_connecting_to_VSC/connecting_to_VSC.html#(21)|How to back up my files]]
+  * Backed up filesystems:
+    * ''%%$HOME%%'' (VSC-3)
+    * ''%%$HOME%%'' (VSC-4)
+    * ''%%$DATA%%'' (VSC-4)
+  * Backups are performed on best effort basis
+    * Full backup run: ~3 days
+  * Backups are used for **disaster recovery only**
+  * Project manager can exclude $DATA filesystem from backup
+    * [[https://service.vsc.ac.at/|service.vsc.ac.at]]
+====== Storage exercises ======
+In these exercises we try to measure the performance of the different storage targets on VSC-3. For that we will use the “IOR” application (https:%%//%%github.com/LLNL/ior) which is a standard benchmark for distributed storage systems.
+“IOR” for these exercises has been built with gcc-4.9 and openmpi-1.10.2 so load these 2 modules first:
+<code>
+module purge
+module load gcc/4.9 openmpi/1.10.2
+</code>
+Now extract the storage exercises to your own Folder.
+<code>
+mkdir my_directory_name
+cd my_directory_name
+cp -r ~training/examples/08_storage_infrastructure/*Benchmark ./
+</code>
+Keep in mind that the results will vary, because there are other users working on the storage targets.
+====== Exercise 1 - Sequential I/O ======
+We will now measure the sequential performance of the different storage targets on VSC-3.
+  - With one process
+<code>
+cd 01_SequentialStorageBenchmark
+# Submit the job
+sbatch 01a_one_process_per_target.slrm
+# Inspect corresponding slurm-*.out files
+</code>
+<HTML><ol start="2" style="list-style-type: lower-alpha;"></HTML>
+<HTML><li></HTML>With 8 processes<HTML></li></HTML><HTML></ol></HTML>
+<code>
+# Submit the job
+sbatch 01b_eight_processes_per_target.slrm
+# Inspect corresponding slurm-*.out files
+</code>
+Take your time and compare the outputs of the 2 different runs. What conclusions can be drawn for the storage targets on VSC-3?
+====== Exercise 1 - Sequential I/O performance discussion ======
+Discuss the following questions with your partner:
+  * The performance of which storage targets improves with the number of processes? Why?
+  * What could you do to further improve the performance of the sequential write throughput? What could be a problem with that?
+  * Bonus Question: ''%%$TMPDIR%%'' seems to scale pretty well with the number of processes although it is an in-memory filesystem. Why is that happening?
+----
+====== Exercise 1 - Sequential I/O performance discussion ======
+Exercise 1a:
+<code>
+HOME: Max Write: 237.37 MiB/sec (248.91 MB/sec)
+GLOBAL: Max Write: 925.64 MiB/sec (970.60 MB/sec)
+BINFL: Max Write: 1859.69 MiB/sec (1950.03 MB/sec)
+BINFS: Max Write: 1065.61 MiB/sec (1117.37 MB/sec)
+TMP: Max Write: 2414.70 MiB/sec (2531.99 MB/sec)
+</code>
+Exercise 1b:
+<code>
+HOME: Max Write: 371.76 MiB/sec (389.82 MB/sec)
+GLOBAL: Max Write: 2195.28 MiB/sec (2301.91 MB/sec)
+BINFL: Max Write: 2895.24 MiB/sec (3035.88 MB/sec)
+BINFS: Max Write: 2950.23 MiB/sec (3093.54 MB/sec)
+TMP: Max Write: 16764.76 MiB/sec (17579.12 MB/sec)
+</code>
+----
+====== Exercise 2 - Random I/O ======
+We will now measure the storage performance for tiny 4kilobyte random writes.
+  - With one process
+<code>
+cd 02_RandomioStorageBenchmark
+# Submit the job
+sbatch 02a_one_process_per_target.slrm
+# Inspect corresponding slurm-*.out files
+</code>
+<HTML><ol start="2" style="list-style-type: lower-alpha;"></HTML>
+<HTML><li></HTML>With 8 processes<HTML></li></HTML><HTML></ol></HTML>
+<code>
+# Submit the job
+sbatch 02b_eight_processes_per_target.slrm
+# Inspect corresponding slurm-*.out files
+</code>
+Take your time and compare the outputs of the 2 different runs. Do additional processes speed up the I/O activity?
+Now compare your Results to the sequential run in exercise 1. What can be concluded for random I/O versus sequential I/O on the VSC-3 storage targets?
+====== Exercise 2 - Random I/O performance discussion ======
+Discuss the following questions with your partner:
+  * Which storage targets on VSC-3 are especially suited for doing random I/O
+  * Which storage targets should never be used for random I/O
+  * You have a program that needs 32Gigabytes of RAM and does heavy random I/O on a 10 Gigabyte file which is stored on ''%%$GLOBAL%%''. How could you speed up your application?
+  * Bonus Question: Why are SSDs so much faster than traditional disks, when it comes to random I/O? (A modern datacenter SSD can deliver ~1000 times more IOPS than a traditional disk)
+----
+====== Exercise 2 - Random I/O performance discussion ======
+Exercise 2a:
+<code>
+HOME: Max Write: 216.06 MiB/sec (226.56 MB/sec)
+GLOBAL: Max Write: 56.13 MiB/sec (58.86 MB/sec)
+BINFL: Max Write: 42.70 MiB/sec (44.77 MB/sec)
+BINFS: Max Write: 41.39 MiB/sec (43.40 MB/sec)
+TMP: Max Write: 1428.41 MiB/sec (1497.80 MB/sec)
+</code>
+Exercise 2b:
+<code>
+HOME: Max Write: 249.11 MiB/sec (261.21 MB/sec)
+GLOBAL: Max Write: 235.51 MiB/sec (246.95 MB/sec)
+BINFL: Max Write: 414.46 MiB/sec (434.59 MB/sec)
+BINFS: Max Write: 431.59 MiB/sec (452.55 MB/sec)
+TMP: Max Write: 10551.71 MiB/sec (11064.27 MB/sec)
+</code>
+----