Cisco released version 2.1(1b) of the HX data platform on 28 april of 2017.
Regarding the UCS firmware:
This version requires Cisco UCS software version 3.1(2g), which was 3.1(2b) with the HX data platform version 2.0(1b).
With this new version, there are no new features but there is a small amount of new hardware support:
Support for the following CPUs: UCS-CPU-E52699AE and UCS-CPU-E78894E
Support for the following new drives: UCS-SD480GBKS-EV and UCS-SD19TBKSS-EV
But more importantly support for the following HX-AF nodes: HXAF220C-M4S (C220 M4) and HXAF240C-M4SX (C240 M4SX)
which raises a few questions:
- Dennis, does this mean that the HX-AF nodes are only supported with version 3.1(2g)?
my answer: The Cisco release notes state that they are supported from version 3.1(2e).
- Dennis, does this mean that the HX-AF nodes are not supported with UCS version 3.1(2b), which was required for HX data platform 2.0(1b) when AF was introduced?
my answer: I don’t have an official statement from Cisco regarding this support-question, but with the introduction of the HX data platform 2.0(1b) the HX-AF were already supported. I will suggest to upgrade the firmware to the supported version as soon as possible (it’s not an huge amount of work with UCS) as it has new benefits which I will discuss below!
The list of resolved issues is long and my recommendation should be that everyone (regardless of Hyperflex) should upgrade to 3.1(2f) or higher.
At the time of writing: 3.1(2f) is founded as the stable version:
Regarding the Hyperflex data platform version:
The list of new features is longer that when version 2.0(1a) was released!
- Self Encrypting Disk (SED’s) are now supported. For those environments that need encryption finally hyperflex will support it.
- Enhanced Call Home support, adds an option for collecting logs through HTTPS.
- Cisco Hyperflex Edge: afaik .. a rather expensive solution for the SMB market. -or- a highly scalable solution for the not needed scalable environments –or– .. ok.. I will stop now .. Cisco, get your horses straight: put your R&D time in more needed features like replication, erasure coding, etc.
- Enhanced Cluster scaling .. yeah!!#n!:
- up to 32 Hyper Converged nodes in one cluster
- up to 16 HCI nodes + 16 compute nodes in one cluster
- Support for adding new or existing UCS-FI domains with the FI 6300 series and VIC1387.
- Optimization available for capacity tier (through TAC support for hybrid clusters).
These optimizations are designed for the AF-clusters. When upgrading an AF-cluster, the optimization will be configured out-of-the-box. For Hybrid cluster without performance problems, these optimizations are not configured out-of-the-box, you have to contact TAC for a customized upgrade path to enable these optimizations.
In my previous blog I talked about the need customized caching algorithms for AF cluster, now those customization are out there!
As my lab doesn’t have AF nodes .. I can’t test those new caching algorithms on a good way.