Ideas for 1.0 release

[sitedesign]

I have now been using PVE in our production rack at the datacentre running on 2 HP rack nodes so far as a cluster, using SAS drives in RAID5.

Hugely impressed with the system, setting-up Ubuntu servers is a breeze. I will keep you posted on our findings and write a nice testimonial for your site.

I have a few ideas for version 1.0 (although you have probably already thought of them) :-

1) Disk management - ability to add extra drives to a node and add them to the drive space or use them for backups etc.

2) Template locations - ability to change the template locations so they can be stored on a separate cheap SATA drive for instance instead of the higher priced SAS RAID space.

3) Template management - ability to set-up a container with a template, add some more packages to it then convert it back into a template for rapid deployment of server configurations.

4) Option to (3) Container copying - ability to take a running container and copy it to another container but change the IP address allocated and other fix-ups to prevent conflicts.

5) Physical Linux server migration script - a script which can be run on a physical Linux server to back it up and convert it into a container file, I know there a few instructions for doing this but they are not very up to date or easy to follow for Debian users. Even better would be to use Rsync on the physical server then point the PVE interface at it for an automated method (not sure if that is even possible).

That's all I can think of so you have clearly done a fantastic job.

Regards to you all

Peter King

Technical Director
Web Sitedesign.net Ltd
www.sitedesign.net

[dietmar]

1) Disk management - ability to add extra drives to a node and add them to the drive space or use them for backups etc.

something like a 'drive space' is planned.

2) Template locations - ability to change the template locations so they can be stored on a separate cheap SATA drive for instance instead of the higher priced SAS RAID space.

Well, most templates are less than 200MB - not sure if its worth.

3) Template management - ability to set-up a container with a template, add some more packages to it then convert it back into a template for rapid deployment of server configurations.

We have some handy makefiles to build templates - maybe we can extend them to provide such functionality.

4) Option to (3) Container copying - ability to take a running container and copy it to another container but change the IP address allocated and other fix-ups to prevent conflicts.

planned.

5) Physical Linux server migration script - a script which can be run on a physical Linux server to back it up and convert it into a container file, I know there a few instructions for doing this but they are not very up to date or easy to follow for Debian users. Even better would be to use Rsync on the physical server then point the PVE interface at it for an automated method (not sure if that is even possible).

this is interesting, but has no priority yet - to many other thing to do.

- Dietmar

[sitedesign]

That sounds excellent, I am very much looking forward to seeing version 1.0.
Are you going to keep this project as open source or will you be making it commercial later?
I would like to see a section in the wiki about the features for version 1.0 so that the community could offer discussions, ideas and help testing.
Do you have any release schedule planned yet, again that would be nice to see on the wiki.
I have noticed that a few of the forum users have started posting howto style replies which would be nice to see a new section in the forum for these type of posts to be moved into. Then as they become more common they can be migrated to the wiki.

Excellent work. PVE has saved us about £12,000 against buying the VMware solution that didn't as good a job!

Regards Peter King

[tom]

That sounds excellent, I am very much looking forward to seeing version 1.0.
Are you going to keep this project as open source or will you be making it commercial later?
I would like to see a section in the wiki about the features for version 1.0 so that the community could offer discussions, ideas and help testing.
Do you have any release schedule planned yet, again that would be nice to see on the wiki.
I have noticed that a few of the forum users have started posting howto style replies which would be nice to see a new section in the forum for these type of posts to be moved into. Then as they become more common they can be migrated to the wiki.

Excellent work. PVE has saved us about £12,000 against buying the VMware solution that didn't as good a job!

Regards Peter King

Hi Peter,

Thanks for your testing and feedback. If you want to speed up development of 1.0 you can think of supporting the Proxmox VE testing labs here by donating a part of your saved money.

We will mention all people and companies supporting ProxmoxVE on the project web page.

[sitedesign]

I will check with our accounts department how I can do that, normally I have to submit an expenditure report to get a budget and then submit invoices for all expenses. I am sure we can get a bit for you though.

[Erwin123]

That sounds excellent, I am very much looking forward to seeing version 1.0.
Are you going to keep this project as open source or will you be making it commercial later?
I would like to see a section in the wiki about the features for version 1.0 so that the community could offer discussions, ideas and help testing.
Do you have any release schedule planned yet, again that would be nice to see on the wiki.
I have noticed that a few of the forum users have started posting howto style replies which would be nice to see a new section in the forum for these type of posts to be moved into. Then as they become more common they can be migrated to the wiki.

Excellent work. PVE has saved us about £12,000 against buying the VMware solution that didn't as good a job!

Regards Peter King

The option to copy containers would save us a LOT of work.

I'm sure proxmox could make a lot of money with ve and still keep it open source. I will even defend the position they will be able to make a lot more money to by keeping the product open source...
it will be the difference between making good money or go huge

[Erwin123]

Are you going to keep this project as open source or will you be making it commercial later?

Dietmar or Tom, can you comment on this?

My worst fear is that you'll adopt a 'monthly paying Parallels like scheme' in wich you have to sell your soul and company to Parallels to use their products.
That was the first reason I started looking for alternatives and stumbled on PVE in the first place.

[dietmar]

My worst fear is that you'll adopt a 'monthly paying Parallels like scheme' in wich you have to sell your soul and company to Parallels to use their products.
That was the first reason I started looking for alternatives and stumbled on PVE in the first place.

The whole project is GPLed, and there are no plans to change that.

- Dietmar

[Erwin123]

I'm currently at a Usenix conference in Boston visiting Xen and Vmware tutorials and lectures. A great way to promote a product.
I hope I'll meet you someday on one of these conferences.

I've said it before but keep repeating it, if there's anything I can do to help/promote/endorse the product let me know.
And I'll gladly donate once I'm back (be it symbolic amount, we don't have budgets mentioned above yet).

[kro]

2) Template locations - ability to change the template locations so they can be stored on a separate cheap SATA drive for instance instead of the higher priced SAS RAID space.

configurable template repository: Configure in a webbased form
-> location of .apl-file (ftp://)
-> gpg pubkey (or keyID)

Would make it easy to integrate an own repository in the create-Dialog.
Same thing would come handy for the .iso files - you can put up one mirror in your network and all cluster members can pull from there.


Just some more thoughts...

[fortechitsolutions]

For what it is worth, if it was not too painful to integrate support of iSCSI initiator (client) to ProxMox, it would be amazing. (ie, not intended to replace local disk as "boot environment and core server" - but rather, as a location for additional storage (either misc storage or possibly even for production VMs which are less IO intensive).

OpenFiler as an iSCSI target is "very viable" presently and makes 'inexpensive commodity iSCSI targets' a reality now. You can get upwards of 100mb/sec throughput to such iSCSI targets .. which may be adequate as a backing store for virtualized system images in some cases :-)

By supporting a 'standards based' storage abstraction layer like iSCSI, it might simplify your life too in developing proxmox ( i.e., let the iSCSI target deal with the fiddley details of LVM / Snapshots / etc.) -- and would provide a pretty robust scaling model too for increased storage at reasonable costs. (of course, clients could choose to use non-free iSCSI targets // bigger commercial iSCSI product targets if they so desire -- but having this option open would .. simply provide some nice options.)




Just some daydreaming :-)


--Tim Chipman

[tom]

hi,

fully agree, see our roadmap

[fortechitsolutions]

Oops - best learn how to read ?! :-)

Thanks - glad to hear this is there...