I'm sure it's a great article...
I'm sure it's a great article but I really can't stop staring at her eyebrows.
Put storage inside the individual hosts of a virtual cluster? You're CRAZY... Like a fox
Hyperconvergence, putting storage inside the individual hosts of a virtual cluster, was supposed to save us from the cost and expense of centralized storage. Thus far, mainstream providers of hyperconvergence have largely failed to deliver on this promise. 2016 looks set to be the year this finally changes. Nutanix has been …
-
-
Thursday 23rd June 2016 11:05 GMT Voland's right hand
Re: I'm sure it's a great article...
Snap out of it, she can be your daughter.
I need to make sure my daughter does not see the pic as that means adding a Fenec to the next birthday pester list.
So while you are staring at the eyebrows and I am staring at the Fenec, who is actually reading the article?
-
-
-
Thursday 23rd June 2016 12:54 GMT Roger Kynaston
Re: who is actually reading the article?
I read it as well but didn't understand it. In the BS bingo league though this seems to score quite highly and hyper is the new term so I propose we hyperconverge our hypervisors with a hypertext interface. In that way we can realise the full potential of our hyperbole.
-
-
Thursday 23rd June 2016 16:14 GMT Anonymous Coward
Re: I'm sure it's a great article...
> Snap out of it, she can be your daughter.
Oh sure. I'm not staring in a pervy way. More in a bewildered "what on earth has she done to her eyebrows; and why on earth would you want to do that" kind of way. Mostly a "what on earth has she done" kind of way.
In know, I know, you're going to tell me that it is "fashion" but I can't help thinking that they are some form of tribute band to the fox.
-
-
-
-
Thursday 23rd June 2016 13:24 GMT Anonymous Coward
Storage in individual hosts is a software problem
Even if you have infinitely fast links, you still have this extra complication. And you have to replicate everything at least once, to account for host failures, and maintenance would be more complicated as you'd have to insure you didn't take down all copies of active data, etc.
If you are already going to have complex software that every host depends on, and need at least 2x the actual storage (probably 3x to be safe) in addition to requiring higher performance from individual host connections, exactly where are the savings over centralized storage?
Even Google, who initially used all that unused space in the hard drives for all their servers have modified that scenario and have dedicated storage nodes now. That should tell us something...
-
Thursday 23rd June 2016 16:35 GMT Ken-in-Houston
Re: Storage in individual hosts is a software problem
Nutanix offers storage-only nodes.
Also, with regard to making things more complicated, HCI actually makes things LESS complicated.
In the absence of a SAN, there is no need to configure LUNs, no need to mask them, manage them, etc. Also, without LUNs, storage access is now much easier and no longer limited: In the past, if you wanted to add storage, you typically had to add an additional LUN. Or, if you were fortunate enough to be able to expand a LUN (not possible with several vendors), that would require downtime to perform.
Nutanix allows expansion of storage w/o downtime. Actually, expansion of storage is w/o configuration, too! It is literally a 2-click operation.
You should take a look at Nutanix first, and then form an opinion. It's obvious you do not understand the technology right now.
-
Monday 27th June 2016 06:16 GMT Lost_Signal
Re: Storage in individual hosts is a software problem
Ken in Houston, expanding LUNs with vSphere VMFS and Windows (2008r2 at least) doesn't require downtime. 10 year old EVAs could do this.
I agree that LUNs suck, but NFS mount points can still cause issues for management. If you don't want LUNs vvols and VSAN with storage based policy management both have you covered (no plugins needed!).
John-In-Houston :)
-
-
-
Thursday 23rd June 2016 13:38 GMT Anonymous Coward
I think many people are way too eager to throw out tried and tested traditional storage. What one should keep in mind is that traditional storage becomes cheaper and cheaper too. So just to give you an example, with Nutanix, you would probably need 2 x 3 nodes, equalling 6 nodes in total, assuming that minimum cluster size is 3 nodes, and that you would like to replicate it to a second site.
On top of that, you need your switch infrastructure, let's assume minimum 2 switches, 10 Gbps, on both sites, so a total of 4.
Now, just for fun, compare the price with 2 ESX servers (or use existing ones), connected with SAS to avoid one switch layer (if possible, do not combine storage traffic and other traffic on the same ethernet switches), and 2 x entry level storage such as the SCv2000, and see where the price ends up.
I'm not sure the price ends up in favor of Nutanix. Of course you then need to factor in cost of administration, but traditional storage has one interface, and nutanix has another interface. On top of that, there's vCenter as well.
So yes, hyper convergence is simple for predictable workloads, but with traditional solutions, you can scale disk, cpu and network independently of each others.
What is best? That's up to the user to decide. =)
-
Thursday 23rd June 2016 20:20 GMT Anonymous Coward
Theyre priming for the other news item :-
http://www.theregister.co.uk/2016/06/23/huawei_to_start_flogging_software_only_next_year/
Part of that telco as a box will include virtual machines running storage aray clusters. Powered by... Huawei storage arrays.
Storage vendors warm up your quotes? no, get ready to be bypassed by a one size fits all solution cutting you out the market in one fell swoop more like.
Biting the hand that feeds IT
