Highlighted
amit_emc
1 Copper

VxRAIL POC Queries

VxRAIL architecture - Physical

1.how servers communicate with drives

2. is there any back pane between drives & servers, if yes one common interface or each server has different ? is this back pane redundant ?


3.What if any Nic of any servers fails is it whole appliance down situation or just one server ? same of rest of the components, which component in single node will lead to complete appliance down ?

4. why we need to have all 8 ports connected be it on a single swithc or 2 switch to make appliance up ?


5. both 10G ports active active or active passive ? if active active then why only one vmnic is enabled ?

6. If host-1 goes down which host the vCenter/PSC/VxRAIL Manager along with Prodcution VM's what happens when HOST-1 goes down ? Only management is unavailable or Production VM's will also wont move ?

7. Can vCenter VM/PSC/VxRAIL Manager will vMOtion in event of host 1 down ?

8. How does Erasure coding differ from RAID, and how  33% space is used in EC1 even when it is as good as RAID-5 ?

9. How do you define no. of IOPS on per system ?


10. Marvin & Loudmouth service use,functionality and dependencies ?

11.Why only IPV6 multicast required for Management traffic ?

12. VxRAIL Manager has graceful shutdown option, will this reboot all host or just the VxRAIL appliance cluster ?

If there is any alert (for ex. drive not accessible) Graceful shutdown doesn't allow to shutdown, then what is way out ?

13. When trying to SCP any VM from customer's external host , if fails due to permission issue , why ?

14. Created a folder under Marvin Data store, SCP VM from external host, if worked, but when tried multiple VM SCP in same folder, it failed " no space left" where as total space available is 9TB , data trying to copy is 2 TB, any resolution to this ?

15. Tried copy 2TB VM , it worked, so we can't copy directly to host, we copy on Marvin data store which has 1 drive (3.84)from each host so how does this copy works, block are written sequentially to all host one by one or all data first goes to node 1 drive 1 and if it fills then move to second node drive 1 ?Jeremy.MerrillVxRail Appliance

0 Kudos
2 Replies

Re: VxRAIL POC Queries

Amit:

I suggest starting with the VxRail tech book - it contains answers to many of these questions. Dell EMC VxRail Appliance Techbook - EMC Corporation and the Netowrking Guide DELL EMC VxRAIL™ NETWORK GUIDE - EMC Corporation both contain answers to many of your questions.

Also, please specify th configuration of the appliance in your PoC - 4 nodes or 3?

For #12 please open a support request

0 Kudos
Jeremy_Merrill
2 Bronze

Re: VxRAIL POC Queries

amit_emc wrote:

VxRAIL architecture - Physical

1.how servers communicate with drives

2. is there any back pane between drives & servers, if yes one common interface or each server has different ? is this back pane redundant ?


3.What if any Nic of any servers fails is it whole appliance down situation or just one server ? same of rest of the components, which component in single node will lead to complete appliance down ?

4. why we need to have all 8 ports connected be it on a single swithc or 2 switch to make appliance up ?


5. both 10G ports active active or active passive ? if active active then why only one vmnic is enabled ?

6. If host-1 goes down which host the vCenter/PSC/VxRAIL Manager along with Prodcution VM's what happens when HOST-1 goes down ? Only management is unavailable or Production VM's will also wont move ?

7. Can vCenter VM/PSC/VxRAIL Manager will vMOtion in event of host 1 down ?

8. How does Erasure coding differ from RAID, and how  33% space is used in EC1 even when it is as good as RAID-5 ?

9. How do you define no. of IOPS on per system ?


10. Marvin & Loudmouth service use,functionality and dependencies ?

11.Why only IPV6 multicast required for Management traffic ?

12. VxRAIL Manager has graceful shutdown option, will this reboot all host or just the VxRAIL appliance cluster ?

If there is any alert (for ex. drive not accessible) Graceful shutdown doesn't allow to shutdown, then what is way out ?

13. When trying to SCP any VM from customer's external host , if fails due to permission issue , why ?

14. Created a folder under Marvin Data store, SCP VM from external host, if worked, but when tried multiple VM SCP in same folder, it failed " no space left" where as total space available is 9TB , data trying to copy is 2 TB, any resolution to this ?

15. Tried copy 2TB VM , it worked, so we can't copy directly to host, we copy on Marvin data store which has 1 drive (3.84)from each host so how does this copy works, block are written sequentially to all host one by one or all data first goes to node 1 drive 1 and if it fills then move to second node drive 1 ?Jeremy.MerrillVxRail Appliance

Amit - I agree with Colin on this.  The 2 guides he pointed you to answer most of those questions.  For #9, the IOPs will really vary based off workload (random, sequential, working set size, R/W Ratio).  It's hard to say a VxRail G Series will give you xx,xxx IOPs because it won't be the same for every customer.  For #13 or #14, there's also a datastore just for the SATADOM on each node...it could be that you created the folder in that datastore.  It could also be that by default ssh is disabled on the ESXi nodes.

I'd suggest taking a look at those documents to see if they answer you questions, and let us know which ones are still open after looking at them.

0 Kudos