Rich15's Posts

Rich15's Posts

Hi, Yes, Change Tracker is fully documented in the Solutions Enabler Doc Array Management Chapter 9.  Essentially you set up a session for a group of Symm devices and define the time intervals... See more...
Hi, Yes, Change Tracker is fully documented in the Solutions Enabler Doc Array Management Chapter 9.  Essentially you set up a session for a group of Symm devices and define the time intervals you want the report to produce as well as how long the monitor should run.  There is no overhead in doing this, essentially its just capturing some metrics.  But have a CT session on a devices, may block some symconfigure some changes so be aware of the fine print.
Hi, The SRDF metrics dont actually show the amount of change, they simply show the traffic rates per second. So effectively the metrics measure measure bandwidth usage.  To get an actual chang... See more...
Hi, The SRDF metrics dont actually show the amount of change, they simply show the traffic rates per second. So effectively the metrics measure measure bandwidth usage.  To get an actual change rate for any devices, you run the Change Tracker tool which is part of SOlutions Enabler. You set that up to monitor the tracks that are updated over given intervals.  This is useful in planning for TimeFInder but really isnt directly related to SRDF. The reason is how "change" is measured.  If you do a write to a track, that track will be sent immediately in SRDF/S mode or delayed in SRDF/A mode.  But that same track may be updated many times before it actually is de-staged to local drives.  Solutions Enabler can measure how many changes to a track took place over an interval. The SRDF metrics will just show the rate of data being sent across the link.  So it really depends on why you need to know the change rate?
Each of the Unisphere sections that can be monitored or configured using the REST API are separated into different resources and detailed in the REST API documents. The documents are available h... See more...
Each of the Unisphere sections that can be monitored or configured using the REST API are separated into different resources and detailed in the REST API documents. The documents are available here:  https://UNIVMAX_IP:UNIVMAX_PORT/univmax/restapi/docs , Where, UNIVMAX_IP is the IP address and UNIVMAX_PORT is the port, of the host running Unisphere for VMAX. Once the Zip file has been downloaded, extract the zip file contents and open “index.html”. All URLs, Required parameters and returned results for each resource are detailed in full within these navigable REST API documents. The subsequent WADL file is also available from this location.
The REST Api is the only way to communicate with the Uni database.  But I would also take a look at why it would be necessary to maintain a separate database outside of Uni?  The Uni DB can hold ... See more...
The REST Api is the only way to communicate with the Uni database.  But I would also take a look at why it would be necessary to maintain a separate database outside of Uni?  The Uni DB can hold 15 months of historical data so its very long term.  You can easily have any report / chart you create exported as a CSV file.  In Uni V8, you can schedule this to be run and emailed to you automatically. So if you simply want some basic metrics like array IOPPS, MBS, etc...go this route.
Hi, The Solutions Enabler release notes for version 8 show that the stpdaemon is no longer supported on any VMAX.  In a Vmax 3, the process of STP does not exist.  The ability to use the metrics... See more...
Hi, The Solutions Enabler release notes for version 8 show that the stpdaemon is no longer supported on any VMAX.  In a Vmax 3, the process of STP does not exist.  The ability to use the metrics we collect now stem from Unisphere itself.  You can extract these using the REST API functionality.  However, I would really want to understand what is the purpose of taking the raw data metrics from TTP externally?  The Unisphere database has a 15 month history.  You can also export a particular screen or report in Uni as a CSV so that may be an option.  In Uni V8, you can schedule this and have  it emailed to you directly.
Hi Can you check out page 38 ? Let me know if you have a question Rich Mobile: 201-410-7894 Disrupt Your World with VMAX Service Level Provisioning
I have done this one many times with customers. The procedure is very well documented and it consists of two parts. First you will upgrade Solutions Enable to the same level version 8. I would ... See more...
I have done this one many times with customers. The procedure is very well documented and it consists of two parts. First you will upgrade Solutions Enable to the same level version 8. I would always suggest making a copy of the symapi_db.bin just in case since that’s the only file that is critical. After upgrading SE, then do a database backup of your Unisphere. Lastly, perform the Unisphere upgrade procedure to get to latest version and restore the database from your backup so you don’t lose any history. It’s a very clean process and works nicely plus you have the backups of the critical files just in case. Rich Mobile: 201-410-7894 Disrupt Your World with VMAX Service Level Provisioning
One of the main benefits is more flexibility in the number of RDF ports.  With the native Gige boards, you are installing in groups of 4, using FC ports, you can have whatever number is most appr... See more...
One of the main benefits is more flexibility in the number of RDF ports.  With the native Gige boards, you are installing in groups of 4, using FC ports, you can have whatever number is most appropriate for the workload.  The second benefit is a higher compression rate than with the native boards.  By offloading the compression and TCP encapsulation work to the network hardware, the FC port achieves a higher throughput than a Gige can.  Lastly, error recovery on less than optimal WAN circuits is handled far better by the network hardware than at the individual Gige port level.  The most solid SRDF replication design always use FCIP with RF fibre ports on a VMAX array. The cost difference usually is very small if any since a small Brocade 7800 isn't expensive and the FC IO modules are cheaper than the Gige modules. I would certainly look into the FCIP design first all the time.
Hi, Keep in mind that the symclone create –vse that needs to be done may take a longer amount of time. During this time, the extents related to the full sized clone are being released back in... See more...
Hi, Keep in mind that the symclone create –vse that needs to be done may take a longer amount of time. During this time, the extents related to the full sized clone are being released back into the pool and marked as free. Depending on the other pool activity this could take several minutes. So I would schedule the create process ahead of the activate process to ensure you have sufficient lead time. This only happens for a create that is essentially converting a standard clone to a VP Snap. Rich Mobile: 201-410-7894
Hi, Given that any report would only be accurate at that moment in time, this is probably not a equitable approach to charging. It is similar to what occurred in VMAX 2 using FAST. Some cust... See more...
Hi, Given that any report would only be accurate at that moment in time, this is probably not a equitable approach to charging. It is similar to what occurred in VMAX 2 using FAST. Some customers took a snapshot once a week of where each SG lived and charged for that type of space. But that had no true indication of the usage over longer periods of time. You could easily be taking the snapshot when a SG wasn’t using much SSD but a day later it does. The most effective and accurate approach to chargeback has always been to use the FAST policy and now in a VMAX 3 to use the SLO. That ties it directly to the service time. You can easily see how much capacity an SG has actually used and charge a certain rate for Gold. If you were billing me, this is exactly how I would want it. Otherwise I would suggest you are taking a snapshot when my usage of the expensive tier is high and it’s not appropriate. Rich Mobile: 201-410-7894
Hi, The challenge is to see the UCB and VOLSER, the SW needs to reside on the MF host.  Any external connection like ECC / Solutions Enabler cannot get that info directly. I am hoping that you... See more...
Hi, The challenge is to see the UCB and VOLSER, the SW needs to reside on the MF host.  Any external connection like ECC / Solutions Enabler cannot get that info directly. I am hoping that your MF host is running the EMC software Mainframe Enablers? That can produce displays of device mappings for you via a batch job.  That output can be captured and placed into a spreadsheet.  Another option is to have the MF host install the Z/OS Solutions Enabler SW (free).  This runs on the MF LPAR and can do that mapping of UCB / VOLSER (for only that LPAR) which can feed open systems management software. I would also suggest upgrading to Unisphere as part of this since it will be easier working with that tool, assuming the Z/OS Solutions Enabler is running and you have ability to communicate with it. I would start down that path first. There is a white paper on using Unisphere to manage and monitor a MF environment available.
Greetings, My customer is pointing out that doing provisioning from Vipr to VMAX, the method is to use a meta member size of 240GB, which is the largest that a VMAX can handle.  Pre-Vipr, we h... See more...
Greetings, My customer is pointing out that doing provisioning from Vipr to VMAX, the method is to use a meta member size of 240GB, which is the largest that a VMAX can handle.  Pre-Vipr, we had always advised using a smaller member size, like 100GB to build meta-volumes.  Since I can enable auto-meta creation within the VMAX to build a member of a certain size, is there any way to alter the default size that Vipr is using ?
Hi, The recommendation is to have both the Data Collection and Data Movement time windows for FAST/VP bet set to 24/7.  There is no good reason to prevent promotion if the host becomes active.... See more...
Hi, The recommendation is to have both the Data Collection and Data Movement time windows for FAST/VP bet set to 24/7.  There is no good reason to prevent promotion if the host becomes active.  You would be introducing performance problems that could have been prevented.
Hi, I think you are thinking about the original FAST or Symm Optimizer.  That was a analyze and move process that happened over several days and moves were planned and approved. FAST VP is ... See more...
Hi, I think you are thinking about the original FAST or Symm Optimizer.  That was a analyze and move process that happened over several days and moves were planned and approved. FAST VP is completely different.  It is a real-time process which is moving small extents all the time that the move time window is open.  You cannot see each of these moves since they are small and dynamic.  The way to see the movement itself is within Unisphere using the Virtual Pools in Performance section.  You will see two metrics: ingress / egress.  Both are tracks per second that are moved by Fast either into or out of that pool.  So for example, the ingress to FC would probably be the egress from SATA as extents get promoted. As shown above, you can always see exactly where a TDEV lives right now, how many tracks in each pool.  FAST VP was never intended to need any customer approval or management nor will it provide extent level history. EMC recommends that the Collection and Movement Time windows for FAST/VP be set to 24/7 since there is no reason to prevent a promotion from happening if the host is active.
Correct, EMC CE needs to do the remaining steps now that you have the physical install completed.  I would also remind the CE to execute the software scripts since they may not always be familiar... See more...
Correct, EMC CE needs to do the remaining steps now that you have the physical install completed.  I would also remind the CE to execute the software scripts since they may not always be familiar with that part.  Its really just the second part of the install and ask if you want to use the mfg layout, then it will execute all the correct CLI. The is an article on the support site describing how to install a 10K but you wont be able to complete it yourself.
Hello, Just to complete the thought, the VMAX 10K is completely configured when it arrives from manufacturing.  Part of the CE installation process is to run a install script.  This is a serie... See more...
Hello, Just to complete the thought, the VMAX 10K is completely configured when it arrives from manufacturing.  Part of the CE installation process is to run a install script.  This is a series of CLI commands which build the TDATS, define the pool,s and define the FAST VP tiers and basic policy.  The customer is not part of this process.  Its completely done by the EMC CE and is part of our installation so no additional charges.  I would strongly recommend that the array be deployed as it was configured since this has proven to be the optimal layout.
This array is 5876.268.145 current target.  I have seen the issue across multiple arrays in 5876 family.
Hello, I have consistently seen in Unisphere 1.6.1.8 that the Host Writes/sec and Host MB Written/sec for an RDF group are always missing. All other metrics are there.  To research further, I... See more...
Hello, I have consistently seen in Unisphere 1.6.1.8 that the Host Writes/sec and Host MB Written/sec for an RDF group are always missing. All other metrics are there.  To research further, I collected STP data from the array and it does have those metrics under teh RDF Group section listed as Kbytes Written per sec.  I cant find this bug in the Unisphere support but I have experienced it on many Unisphere installations.  No mention of anything being corrected in 1.6.2 release notes.  Does anyone know if this is a known issue and will be corrected?  The purpose is to show the write activity generated by the Host side for the group and compare to the RDF transmission value to see how closely they align. Here is a screen shot of the issue:
The array is a pair of VMAX 20K with Gige boards for RDF connectivity.  Noticed that compression is not enabled on the RDF group.  The question is the proper procedure to enable compression. Sh... See more...
The array is a pair of VMAX 20K with Gige boards for RDF connectivity.  Noticed that compression is not enabled on the RDF group.  The question is the proper procedure to enable compression. Should this be done using symconfigure on the GigE RDF port first and then enable it on the RDF group using symrdf? Enable both HW and SW levels? Impact of doing this while SRDFA is running ? Thanks !
No, vmax was never designed as a compliance type array. There is no way make a Lun read only and guarantee it will not be updated. A symm device can be set to read only status but can just as... See more...
No, vmax was never designed as a compliance type array. There is no way make a Lun read only and guarantee it will not be updated. A symm device can be set to read only status but can just as easily set back to read write. So no real security to prevent updating. Rich Sent from my iPad