Hi, I'm Dean Pompilio. Welcome to Cyber Harry will be doing Module six, Lesson five now working with V M. F s data stores. So this lesson will look at what's involved in creating of the NFS datastore
also what's involved in increasing its size
and then, uh, deleting or on mounting a day of store wrapping that up by doing lab number eight,
we created the M. F s data store.
First thing we have to think about is some of the capabilities of the day of store. What? It's best that it is optimized for storing very large files and accessing those files,
we can have a maximum volume size up to 64 terabytes. That's pretty generous.
Ah, and compared to NFS, V. M. F s offers several features out of fast. Does not
NFS is more of a general purpose file system. Raz v. M. F s is
is designed for virtual machine operating environments. So when you create the their store, we go to our configuration tab on the host. Click your storage link.
I'm on the left side and in the upper right side of the of the interface, you'll see the ad storage link.
So that point you just select
their storage type that you like, whether it's a disk or lung
And if we pick a London, we choose from the available lungs, which you'll get to see in the lab.
Then you specify a date, a store name and specify the size,
and it gets created within a couple of seconds usually and now that storage becomes available to your host.
So viewing the data stores once that's been created very simple as well. We go to the configuration tab stores link yet again,
and I should see a list of whatever I have attached.
So I'm gonna have another fast partition listed or some local disc.
In the case of a local disc,
I know what the status is, what devices running on. If it's local disc, I'll know that it's non and non SSD.
I get my capacity information, and then it tells me the bm fast version number
in this case to be in my office five, which is what you should be using as well. If you're on,
he has ***. I version five got five.
In order to browse the data store, you just right click and select browse from the drop down menu,
and the same thing applies to deleting a day, a store or a kn mounting it.
If you delete a day of store, you're actually are deleting
the entire partition.
and in a fast partition, where if you delete or announce the same thing in this case when you delete, it really means delete. So be careful.
If I un mountain, it's just no longer accessible,
and I could re mount at a later time.
Uh, and you could do that just by scanning your your host bust adapters
and the story just show up again.
So some of the things that happened when you're dealing with data stores
are due to the disc provisioning method of the virtual machines that exist on that data store.
We talked a little bit about thin provisioning
in an earlier lecture and how they allowed you to over commit your storage requirements.
I could have a data store that has a capacity of 30 gigabytes,
but I could actually provisions in
enough storage to say that Let's see, I have four V m, each with 10 gigabytes off thin, provisioned desk.
At some point, I'll need 40 gigabytes of storage to hold all of that data,
even though they're there on Lee at one gig about each in the meantime,
via more allows you to over commit your story so I could create four B EMS
on this partition 10 gigabytes each, even though it only has 23 available. It'll still let me do that as long as the disk is thin provisioned,
so eventually that been provisioned, disc will grow to the point where now the day of store might fill up. And, of course, when that happens, you're vey EMS that are, ah,
that on that data store majors freeze or crash because they can't use any more storage and they become unstable and will crash.
So the idea is that if you if you're in this habit of over committing your storage, you have to make sure you keep an eye on the data store.
You could just look at it every day and monitor that way or smarter approach would be to configure some alarms.
We'll cover alarms and later, a lecture and some labs.
The alarm basically lets you set some thresholds to get a warning or too sudden alert when you're nearing a
some condition or an event that's happening. It could be related to storage as we're talking about here, or could be related to remember usage or CPU usage
If you want to be proactive, you should increase the capacity before it's actually required.
So you're watching it. You're watching it. You see you're getting close now. I need to do something about that so I could just add space to the day store, make it larger than it was. And now I I've averted that problem.
Another option, when you're especially if you're in a production environment and
you can't afford to have any downtime, is to move some of the VM somewhere else.
Two other storage. So use your V Motion
in order to relocate those bm files, too.
Another storage area
that might be a better option than some of these other options, depending on what your circumstances are.
All right. So speaking of adding space to a data store or increasing its size
effectively, what we're doing is adding an extent, which is alone logical unit number.
Once you add the extent
then you can just expand the day of store. Within the new extent, this is the same way that it works in a window's environment. For instance, you could expand depending on what version of Windows or Windows server that you have. You could expand your hard drive size
and your let's say your hard drives 10 gig
and you expanded to 20 gigabytes. The file system is still 10 gigabytes. Same thing in a V M F has context.
So we have to expand the file system to now fill that extra 10 gigabytes of space. So we have the full use of that.
Unfortunately, you cannot shrink it in a store.
So once you expand it, once you add extends,
you're stuck with that. The only way to make a smaller data store would be to create a new one
with the size that you require, and then move all this data to the new, smaller data store, and then you can get rid of the old one. That's really no way to test to handle that properly.
When you're increasing the size of Dan store, you can have your VMS powered on. That's a powerful feature,
so back to this situation here where we have overcommitted our storage. I don't have to worry about taking of'em down or having having it off line for any reason. I could just increase the size while it's up and running.
Of course, depending on what the V M is doing, there may be some performance impact
the day of store operations are going on. But it shouldn't be too horrible. Under most circumstances.
Your storage administrator can add lungs
to your ah, to your host or to be a visible to your host,
and the storage administrator can also increase the size of existing ones.
In either case, if Lunger added or their increased,
it's still need to expand the data store
into those new extents once you
apply them to the existing data store.
So there's nothing really automatic about. You still have to take those discreet steps in order to make this happen.
So some of the limitations you can have up to 32 lungs per data store.
Each of those lungs could be up to 64 terabytes each,
or you can have a total of 64 terabytes for the day store as a whole.
So if you want to do the math of different way, I could have 32 2 gigabyte lungs. I could have 1 64 terabyte long. I could have 4 16 terabyte Lund's You get the idea.
The last thing we talk about here is some of the algorithms that are available for doing multi path.
We saw several examples of how multi path thing works. I've got multiple physical paths, either through Fibre Channel or through multiple network interface cards to give me different physical passed through the network for the data to two cards on the host. Two cards on the storage device minimum.
Some storage arrays have their own,
multi patting algorithms, which may be an improvement over the ones that GM where offers.
But you do get three choices
from within Vienna Vienna, where
for third party plug ins there is the applicable storage architecture, or P S. A.
That also works with the native plug ins.
And the idea here is that, for instance, if I had
two paths to adaptors on the host to adaptors on storage array,
I could keep them in an active active mode. That means both of the adaptors are sending data
at the same time or simultaneously more or less the alternate, depending on what
what algorithm we pick.
Or you can go on an active passive mode where I'm always using one adapter until it fails. Then the other one takes over automatically.
So depending on your situation, you might choose. Choose that differently.
is a great multi packing algorithm. Basically, that says that every single time I send any data out of Interface one. I alternate to interface to the next time, and you just go back and forth. If I have three interfaces, I would just go 123123 and so on.
So this lets you get a lot of scalability.
I can add adapters. I can add more storage and grow my capabilities
and have the multi path thing outer of them Keep pace with that
most recently used or M are you
better for providing higher availability. It's not high availability in the sense of a V m, really, but it's something similar. So if I have,
uh, let's go back to the three adapter example.
If a particular host is accessing the storage of on the land, and it's always using
adapter to, and that's the one it will tend to stick with.
If for some reason the future uses it after one, that's the one that will use from that point forward until the host reboots or some other interruption changes that.
this doesn't offer the load bouncing
distribution of trapping the same way that round Robin does. But it does give you the ability to,
uh, spread the load out somewhat between the available adapters.
And then the last option is fixed, which means that by default, when you attach a storage adapter, it will
and a fixed mode, meaning it doesn't switch to any of the other adapters. It stays on the one that's configured until you manually change that to something else.
Okay, so we talked about our
data stores, how they get created
and some of the different characteristics of the volume, size
adding the lungs and the extents were actually works.
We talked a little bit about how to view the parameters everyday store. You can right click to browse, right click to delete, which will delete your files. Remember that
or right click to just on MT.
A little bit about howto manage overcommitted data stores will be careful with your thin provisioning so that you don't over commit yourself
and then, lastly, rediscovered
increase the size of a day store and a little bit about the multi path thing algorithms.
So the next task is to do lab number eight.
So in this lab you'll be seeing how to create the day store, how to change its name
and how to expand it, and then ultimately extended and remove their store,
including lesson Thank you.