Question: What RAID configurations do PS Series (EqualLogic) arrays support?

Answer: The currently supported RAID configurations are: RAID10, RAID10 (no spares), RAID50, RAID50 (no spares), RAID5*, RAID6, RAID6 (no spares), RAID6 Accelerated (only RAID level for hybrid arrays).

 

*Note that RAID 5 does not offer optimal data protection for business-critical data due to the higher risks of encountering a second drive failure during a rebuild, and is therefore not recommended. You cannot configure RAID 5 using the Group Manager GUI; however, it can be configured using the CLI.

  

Question: If we have multiple enclosures in a single SAN, do all of the enclosures need to use the same raid type?

Answer: Each enclosure can be configured with its own RAID type. Each member can only have one RAID type. You can’t select individual drives to create different RAID levels inside one enclosure.

  

Question: Can pools span arrays?

Answer: Yes. A pool may have multiple member arrays. By default the space will be striped across them proportionally by their respective sizes.

  

Question: How is newly released functionality/software delivered to the customer? Can they do or is that a part of maintenance? Is there a cost to new features that come out after the purchase?

Answer: FW updates are made available via the support site. In order to receive FW updates, the array must be under the original one year warranty or have a valid support contract on each array purchased.

  

Question: What software is pre-loaded on the array?

Answer: The array is pre-loaded with the Firmware operating system. FW updates and patches can be downloaded from the support site: eqlsupport.dell.com

  

Question: What additional PS Series software is available for the array?

Answer: Currently the following is available for download and installation to array owners with a valid support contract (or still under warranty): FW Updates; MTU (Manual Transfer Utility); Dell Storage Update Manager (DSUM); SAN HQ (SAN Headquarters); HIT/ME (Host Integration Tools for Microsoft); HIT/LE (Host Integration Tools for Linux); VSM (Storage Replication Adapter for VMware); SRM (VMware Site Recovery Manager); MEM (Multipathing Extension Module for VMware); vCOPS (Dell EqualLogic Adapter for VMware vCenter Operations Manager); Visio Stencils and links to third party iSCSI initiators; and links to SCOM (Dell EqualLogic Storage Management Pack Suite)

  

Question: Is it a PS Series best practice to recommend iSCSI HBAs?

Answer: HW iSCSI HBAs are not a requirement for PS Series arrays. The solution fully supports Ethernet NICs, TOE cards, or iSCSI spec compatible HBAs. HBAs offer boot-from-SAN capability, along with potentially reducing the host CPU workload.  

 

Question: Are Jumbo Frames required?

Answer: No. The PS Series array will auto negotiate standard frames on a per iSCSI session basis.   Jumbo Frames improve network efficiency and should be used if the switch infrastructure supports it. Even if the host does not support Jumbo Frames, inter-array communication will benefit from them.

 

Question: Can I disable Jumbo Frames on the array?

Answer: No. You cannot change the MTU size on the array. However, disabling Jumbo Frames on the switch or host will have the desired effect. The array will always show an MTU size of 9000. At connection time, the MTU size will be negotiated for that session.

 

Question: Is link-level flowcontrol required?

Answer: Yes, in non DCB configurations, all the array and server ports must be configured for flowcontrol.  

 

Question: Should I hard wire the speed and duplex on switch ports?

Answer: No. Auto negotiation is preferred.

  

Question: How is the back end handled in a multi-chassis environment? Is there a master/slave relationship to keep track of volume metadata? What happens if access to the primary array with metadata goes down?

Answer: There is no primary array. They are all "peers." When you have multiple arrays in the same pool, data is striped between members, up to three members by default, or more if the volume size will not fit on three members. In that case, if you lose any member with data for a given set of volumes, those volumes will go offline until the member is restored.

 

Question: Can any of the hard drives be set up as a hot spare?

Answer: Depending on the RAID level it could be one, two or optionally none

 

Question: Does the array do global hot sparing or only in the chassis they are in?

Answer: Hot spares are only available to the chassis they are in.

 

Question: Where is the ACL (access control list) kept? Inside the server? In the MPIO software? How does the server access (R/W) a specific block?

Answer: ACL is on the array. 

 

Question: I'm not getting it. If all arrays are "peers" and an iSCSI is a point-to-point connection, how can the server access the right array that has the right part of its data?

Answer: The host maps to the volume through access control lists for that volume and volume data is spanned across the arrays in a pool. When the server connects to a physical array port, and there is data on another member for that volume, a hidden “MESH” connection is created to the other member(s). Data to or from that member will use that inter-member MESH connection.  

 

Question: It's hard to understand how an I/O (either Read or Write) is fulfilled. Which array responds to a server request, if many might have parts of that volume?

Answer: The array on which the data resides responds to the request.  

 

Question: What about in a Read/Write operation? Which array gets the request?

Answer: If the HIT ME/LE or MEM for ESXi is installed onto the host, all Write (and Read) operations are delivered from the member containing the data, MEM/HIT allows a connection to all the members with data for that volume, so the requests are directly routed to the correct member.

 

Without HIT/MEM installed, if the server has established and connects to member A, and lf the data is on another member, a hidden connection, known as the “MESH” connection, between members is used to get that data on the other member, then the data pass it on to the server via that established iSCSI connection. That same MESH connection will be used to send data to the member that needs it.