Unsolved
This post is more than 5 years old
2 Intern
•
308 Posts
0
1756
Summary for ATE You can ask anything about XtremIO
Summary for ATE
You can ask anything about XtremIO
This is the quick summary for ATE event
【Ask The Expert】エキスパートに聞こう!なんでもこいXtremIO
Detailed Information
【Planning】
Question 2:
About management network support for XtremIO
Answer 2:
NAT environment also supported network between XMS and Storage Controller.
Additional info :How to check network
*Can you ping from Client to storage controller?
if not ....check packet filter /port closure
Question 3:
Hardware issue in XMS server can be detected ?
Answer 3:
Internal drive issue can be detected. But XMS is online or not can only checked by ESRS heart beat every 30 sec via Gateway server (same as VNX CS and VMAX SP )
If there is no response from XMS server,the error ``Device Missing `` occurred and SR become Sev1.
Supported XtremIO configuration need to have ESRS.( In terms of maintenance policy)
Question 9:
XtremIO with Windows, VMware and XtremIO
Answer 9:
Good to have XtremIO Window VSS Provider and VSI for VMware vSphere Web Client for operational purpose.
+VSS Hardware Provider+
See detail on Host Configuration Guide.(https://support.emc.com/docu56210_XtremIO-Host-Configuration-Guide.pdf?language=en_US)
Useful for back up/restore operation.( Some cases are must tool)
+VSI for VMware vSphere Web Client+
Be able to use for function vSphere GUI can perform.( Creating Data store /Provisioning RDM/Action specific to XtremIO vol ) Recommended to install.
See VSI for Web Client Product Guide
https://support.emc.com/docu62318_VSI-for-VMware-vSphere-Web-Client-Version-6.7-Product-Guide.pdf
Question 11:
Overhead of XtremIO against Deduplication
Answer 11:
There is no Performance impact (XtremIO cannot stop Dedup as it is always working inline, Also the more host has dedupdata, the more performance will be)
+Point +
1. Deduplication done inline. Duplicated data is not write to XtremIO,( Nor in chache/SSD )then returns ACK to host.
2.All X-Blick CPU are worked for IO operation evenly.
3. XtremIO will have All Meta data @ inmemory
※ All Other Flash array are usually have metadata in SSD so they cannot perform constant performance .
Performance chat for XtremIO (1:1=NO duplication data 、2:1=1/2 data is duplicated 、10:1=9/10 data are duplicated)
Question 12:
XtremIO Performance if there is no deduplicated data
Answer 12:
Catalog spec per 1 XtremIO X-Brick
8KB Read:Write=70%:30% 150,000 IOPS(Read 100%-> 250,000 IOPS)
condition:
*Data duplication radio (1:1)=All data will be written to SSD
*80% full in array physical capacity
*Response time within 1ms
・The more host has Duplicated data , XtremIO will have better performance
・XtremIO will have stable performance ( not affected by Disk usage ,load etc ..)
Doc :https://www.emc.com/collateral/data-sheet/h12451-xtremio-4-system-specifications-ss.pdf
Question 18:
Pros and Cons about XtremIO data protection
Answer 18:
XtremIO Data Protection(XDP)
Detailed XDP Doc:https://japan.emc.com/collateral/white-paper/h13036-wp-xtremio-data-protection.pdf
PROS>
・High Availability
XDP protects 25 SSD with 23+2 ( raid 6 like double parity ).
Difference between RAID-6 is XDP has diagonally protection so that shorten the calculation time for rebuild.
Data Protection Group(DPG) supports up to 2 disk failure ( same as RAID 6 ) ).
Also high speed rebuilding allow another disk failure during rebuilding .6 SSD issue can be handled per DPG .
・High performance
Overhead ( average upon write operation ) = 1.22
XDP stripe has 8kb data block ( 25 X28 ) and XDP is full stripe.
Related info>
Classical Raid overhead
・RAID-1->2
(1 write need to have one Mirror , means total write block end up 2 . overhead 2 ÷ 1 = 2)
・RAID-5-> 2
(One write need to have one parity …2 ÷ 1 = 2)
・RAID-6->3
Question 21:
The reason for XtremIO has 30% extra effective capacity compare with other flash storage
・Small over head on XDP capacity
Raid technology XtremIO ( 23+2 ) uses less overhead than normal RAID technology .
・No dedicated space for Deduplication and compression ( XtremIO uses inline and in-memory operation )
・No need for garbage collection space
・Effective Capacity is 75% of physical capacity . (Remaining are for system/hot spare/XDP overhead )
eg)10TB model (400GB eMLC SSD x25)-> 7.587TB is Effective Capacity
Question 22:
Data migration to XtremeIO
Answer 22:
If source storage are VNX/VMAX, SAN COPY and Open Replicator can be used.
If Service can be off line, VPLEX also can be used(Not for VNX/VMAX).
Other options are Storage vMotion(vSphere)/Live Migration(MS Hyper-V)/PowerPath Migration Enabler etc .
Question 24:
Sizing for XtremIO
Answer 24:
XtremIO place data by 8 KB. Deduplication rate will be affected how much data duplicated contains per 8 KB.
So host system like virtual environment/VDI which has same OS size line up will have higher rate of deduplication.
Example of Data deduction rate >(deduplication & compression)( Based on actual data )
Database environment: 2-4 times
Virtual server :3-8 times
VDI ( full clone ) server :6-12 times
Useful tool to calculate rate : https://app.mitrend.com/#instructions/XtremIO_Reduction
Question 25 :
MiTrend Calculation
Answer 25 :
1、Install Data Analyzer tool into host and scan by volume
2、result in output.mitrend
3、Upload above file at mitrend.com then analysis report will generated in xcelor ppt
・Scanned data deduction rate
Total |
3.82TB |
Scanned data |
3.82TB |
Necessary Capacity |
0.599TB |
Data deduction rate |
6.52times |
Data deduction from Deduplication and Compression |
6.05 times |
Data deduction from Deduplication |
3.09 times |
Data deduction from Compression |
1.96 times |
・Data deduplcationrate
Above result shows source data 3.82 TB became 599 GB in xtremIO ( deduplication/thin provisioning /compression @ xtremIO )
Question 26 :
Sizing for VDI
Answer 26:
Linked Clone and full clone may not be same size due to deduplication.
【Implementation 】
Question 10:
Adding Hardware to XtremIO
Answer 10:
Brick expansion>
By who: OCE(Online Cluster Expansion) need to be performed by EMC Level 3 engineer.
Impact : Canbe done online.It may affect host performance during data reclamation.
Parameter/setting change : There is no specific change . New Brick storage controller IP only needed.
Question 14:
About multi Array management (Multiple cluster can be managed by One XMS server , Support from XtremIO 4.0~ )
Answer 14:
How to Creating new multiple cluster in one XMS server ?
Create Cluster using create-cluster cmd on XMS server by cluster
Adding new cluster to existing cluster environment or move some cluster into centralized XMS server >
Create new cluster @target XMS server for new cluster ,
Delete cluster registration @ source XMS server then add cluster @ target cluster for existing cluster.
See XtremIO User Guide P.130 "Cluster Migration"
https://support.emc.com/docu62760_XtremIO-4.0.2-Storage-Array-User-Guide.pdf
Question 19:
The reason for no performance tuning XtremIO .
Answer 19:
There is no cache concept as SSD build as single pool.
User only needs to decide Volume size /name because XtremIO using thin volume, All controller are active .Host can be the bottle neck ( see https://support.emc.com/docu56210_XtremIO-Host-Configuration-Guide.pdf)
Question 20:
Tuning opportunity for host /switch
Answer 20 :
Check Host Configuration Guide (https://support.emc.com/docu56210_XtremIO-Host-Configuration-Guide.pdf?language=en_US )
*Page 50 Concurrent full clone operation
*Page 53 VAAI X-Copy tuning
【Operations】
Question 1:
Can user chose specific vol-id upon volume creation?
Answer 1:
vol-id assigned as auto thus user cannot chose it Initiator or object within cluster are the same.
<Volime identitification @ XtremIO>
--Volume name and vol-id cannot be Same. XtremIO/XMS can check from Volume name or Vol-ID.
Question 4:
How to check XMS server internal drive.
Answer 4:
cmd: show-xms-info can be used for usage of CPU/RAM, and usage of Disk.
To check drive failure need to be detected from event or mail alert.
Question 5:
Reporting function for XtremIO ( v4.0~)
Answer 5:
v4.0.x~ Available reports listed below
+Duration and granularity+
Newest 5 sec to 3 days at oldest⇒After 3 collect for 7 days
2 Years at Max ( Data will be deleted After 2 year=720 days )
Granularity Days
5 sec 3 days
1 min 7 days
10 min 30 days
1 h 90 days
1 day 730 days ( 2 year )
+Available Data+
Granularity |
Days |
5 sec |
3 days |
1 min |
7 days |
10 min |
30 days |
1 h |
90 days |
1 day |
730 days ( 2 year ) |
Note:Prior to 4.0, Data can only back up to Up to 1 week ( every 5 sec )with XtremIO cluster basic info ( IOPS, Bandwidth, Latency)
Question 6:
How to Expert performance Data
Answer 6:
CLI(ssh)
Eg)IOPS info every 5 sec ( granunarity=raw) from 23rd Feb 2016. Export to file name iops-cluster.csv
1. Access link for CSV file
2. run cmd ..
xmcli (tech)> show-reports-data category=IOPS entity=Cluster granularity=raw from-time="2016-02-23"
export-to-file="iops_cluster.csv"
Exported query to https:// / xtremapp/iops_cluster.csv
xmcli (tech)>
Question 7:
Stop-cluster command
Answer 7:
Issuing stop-cluster command will stop access to XtremIO volume
Question 8:
About Event notification ( email,SNMP,syslog)
Answer 8:
Usually notifications are set to severity level Major or above.
See detailed thread :https://community.emc.com/message/921920
Question 13:
How to set ESRS
Answer 13:
see below links
333578 : XtremIO: How to Configure XtremIO with ESRS Virtual Edition (ESRS VE)
Question 15:
XtremIO NDU
Answer 15 :
Same as XMS/SC software package and installation,But operation need to be done by EMC support engineer.
Question 17:
XtremIO SNMP v3 MIB download
Answer 17:
You can use v1,v2 MIB file as well. v3 has backward compatible. https://support.emc.com/search/?text=mib&product_id=31111&searchLang=en_US&adv=y&facetResource=DOWN
( https://emcservice.my.salesforce.com/kA2j0000000QXD5CAO
Question23:
How to stop XtremIO
Answer23:
below command
power-off cluster-id=Cluster
shut-down-reason="xxxxx"