Hello,

We are a user of EMC Centera.  We have 8TB raw space.

I have a couple of questions -

1) Why is it that we have only 2.7TB usable?

2) We just got an email from our EMC rep that when the system reaches
85% utilization, we will start having performance problems.  Any
validity to this?  What this (#2) implies is that we actually only have
2.3TB before we run into the performance problem.  Is this just a
marketing ploy to force me to buy more?

Thanks.

[ Post a follow-up to this message ]

Just to be sure I understand what you have, you have a 4-node Centera with
500GB drives installed, correct?

If this is true, you should have 3.4TB of usable space if you are using
mirroring(CPM). Parity(CPP) should yield about 5.8TB usable. If you don't
see this, something is wrong and I would have EMC look at this.

Regards,

Rick
"Mahesh" <mahesh.tailor@gmail.com> wrote in message
news:1149690235.156361.178030@j55g2000cwa.googlegroups.com...
> Hello,
>
> We are a user of EMC Centera.  We have 8TB raw space.
>
> I have a couple of questions -
>
> 1) Why is it that we have only 2.7TB usable?
>
> 2) We just got an email from our EMC rep that when the system reaches
> 85% utilization, we will start having performance problems.  Any
> validity to this?  What this (#2) implies is that we actually only have
> 2.3TB before we run into the performance problem.  Is this just a
> marketing ploy to force me to buy more?
>
> Thanks.
>

[ Post a follow-up to this message ]

There is lot's of 'overhead' with Centera.  Reserved space for rebuild
(reserved regeneration buffers), OS/CentraStar/db spaces etc.

There shouldn't be performance issues at > 85% though, never heard
that.  But, things change w/ different vers, and maybe?  I'd recommend
fwding the comments to your Centera SE - they should have pretty good
lines into performance information.  The next question would be - how
bad a degradation, and where (read/write...) for your environment type
(big files/lots of x activity...).

Issue with operating at higher utilization percentages is usually a
concern over what will happen with the automatic regens, should you
have a disk/node failure and the regen 'fills' the box.

The CLI commands:

show capacity availability
show capacity total
show capacity detail

are pretty good at helping you see where your storage is going.

..clark
clark@storageswitch.com


Mahesh wrote:
> Hello,
>
> We are a user of EMC Centera.  We have 8TB raw space.
>
> I have a couple of questions -
>
> 1) Why is it that we have only 2.7TB usable?
>
> 2) We just got an email from our EMC rep that when the system reaches
> 85% utilization, we will start having performance problems.  Any
> validity to this?  What this (#2) implies is that we actually only have
> 2.3TB before we run into the performance problem.  Is this just a
> marketing ploy to force me to buy more?
>
> Thanks.

[ Post a follow-up to this message ]

Note that you have to have at least 7 storage 'roled' nodes for CPP.
Less than that and they are always CPM.  Beyond that, parity
calculations are really tough, as (for a variety of reasons), lot's of
data still ends up with mirrored (CPM) protection.

What's the regeneration buffer set to?  That'll have the biggest impact
over user controllable space.  May have to have EMC service come reset
this - if you don't like the configuration that you are at.

cli command:
show capacity regenerationbuffer


..clark
clark@storageswitch.com


Rick wrote:[vbcol=seagreen]
> Just to be sure I understand what you have, you have a 4-node Centera with
> 500GB drives installed, correct?
>
> If this is true, you should have 3.4TB of usable space if you are using
> mirroring(CPM). Parity(CPP) should yield about 5.8TB usable. If you don't
> see this, something is wrong and I would have EMC look at this.
>
> Regards,
>
> Rick
> "Mahesh" <mahesh.tailor@gmail.com> wrote in message
> news:1149690235.156361.178030@j55g2000cwa.googlegroups.com... 

[ Post a follow-up to this message ]

> There shouldn't be performance issues at > 85% though, never heard
> that.

Hmm... may be a misunderstanding here.  There are not any performance
concerns related to disk storage, but there *are* performance concerns
as the maximum *object count* approaches.

Aaron

[ Post a follow-up to this message ]