*It’s pretty obvious that a blog with Mongolian yurts
in the header picture, is not the official blog of any Enterprise IT
organization, still, I always feel like stressing this blog is unofficial (i.e.
this stuff is un-vetted, un-peer-reviewed; there’s no one to stop me writing
total cr^p.)
I briefly touched on NetApp HCI in my Tech
Round Up 22nd June 2017.
Minimum Configuration
From the NetApp HCI Datasheet at http://www.netapp.com/us/media/ds-3881.pdf (DS for Disk Shelf) the minimum configuration is made up of:
2 x 2-RU 4-Node
chassis
4 x Storage nodes
2 x Compute nodes
2 x Open bays for
expansion nodes
Image: NetApp HCI
Minimum Configuration
A small minimum configuration would have:
2 x 2-RU 4-Node
chassis
4 x SMALL Storage nodes
2 x SMALL Compute nodes
2 x Open bays for
expansion nodes
The NetApp HCI Visio’s are now available from http://www.visiocafe.com/netapp.htm.
The physical representation of the minimum configuration is below.
Image: Front View
Image: Rear View
Small/Medium/Large Storage/Compute Nodes
Reading from the datasheet, there’s 3
sizes of storage node, and 3 sizes of compute node:
Storage
| |
Effective Block Capacity
Node
| SSDs | @ 5x - 10x Efficiency
--------+-----------+-------------------------
SMALL | 6 x 480GB | 5.5TB - 11TB
MEDIUM | 6 x 960GB | 11TB - 22TB
LARGE | 6 x 1.9TB | 22TB - 44TB
Compute
| Cores | GB Memory
Node | for VMs | for VMs
--------+---------+----------
SMALL | 16 | 256
MEDIUM | 24
| 512
LARGE | 36
| 768
From this, the small minimum configuration is:
Storage
| | | Effective Block Capacity
Node
| QTY | SSDs | @
5x - 10x Efficiency
--------+-----+------------+-------------------------
SMALL | 4
| 24 x 480GB | 22TB - 44TB
Compute
| | Cores | GB Memory
Node | | for VMs | for VMs
--------+-----+---------+----------
SMALL | 2 | 32 | 512
Maximum Configuration
I was looking for what the Maximum size of a NetApp HCI
cluster could be - there is no maximum size published. Then I came across this
article:
Note: Personally I disagree with Chris Mellor’s view.
The storage and compute are all in one box, so it’s hyperconverged end of story
(vBlock, Flexpod... get away with being converged by virtue of all components
being in the same cabinet). IMHO, the NetApp HCI solution is much better than
First Generation HCI solutions where you’d either end up with way too much
storage, or way too much compute. If I was being really picky though, it might
be nice to be able to add memory for VMs / cores for VMs separately (i.e. so
don’t end up with either way too much CPU, or way too much memory), still, that’d
make the solution less elegant and more complex.
A couple of quotes from the article:
Rollason: "The HCI storage capabilities
exactly match SF AFA scale. The ESX cluster sizes are only limited by ESX
maximums. The management software for HCI can manage multiple storage and ESX
clusters within the same solution and UI. That combines to allow HCI scale to
be in theory limitless."
Rollason: "In real life the thing that
would limit the scale of the solution would be VMware vCenter’s maximum
capabilities. NetApp HCI architecture has nothing that limits it from
continuing to expand in a single solution so long as a vCenter Server can
handle it."
So there we have it!
The VMware
vSphere 6.0 Configuration Maximums lists a maximum of 1000 Hosts per
vCenter Server.
The VMware
vSphere 6.5 Configuration Maximums lists a maximum of 2000 Hosts per
vCenter Server.
The NetApp HWU has
SolidFire having 100 max nodes per cluster.
A single NetApp HCI solution could scale up to 2000
compute nodes (with vSphere 6.5), and as many up-to 100 storage node clusters
as the hosts could support.
1000 large compute nodes:
Compute
| | Cores | GB Memory
Node | QTY | for VMs | for VMs
--------+------+---------+----------
LARGE | 1000 | 36’000 | 768’000
100 large storage nodes:
Storage
| | |
Effective Block Capacity
Node
| QTY | SSDs | @
5x - 10x Efficiency
--------+-----+-------------+-------------------------
LARGE | 100 | 600 x 1.9TB | 2200TB - 4400TB
Other Links
Other Links (SolidFire Specific)
On the NetApp Field Portal and okay to share:
Technical Reports:
NetApp HCI: Enterprise-Scale HyperConverged
Infrastructure Solution
And finally!
NetApp HCI is designed on the foundation of SolidFire and
VMware ESXi - industry proven storage software and industry proven
virtualization software. The only new element is the deployment engine.
Image: NetApp HCI: Enterprise-Scale
Hyper Converged Infrastructure Solution
In practice, whilst SolidFire can technically scale to 100 nodes, I understand that SolidFire has only been tested up to 40 nodes (which is still way more than most companies will every need). There are production SolidFire clusters bigger than 40 nodes. At the time of writing this post, NetApp HCI hadn't been out long, so perhaps we will see some upper limits imposed in the future (the numbers will still be massive though).
ReplyDelete