Time line for road map?

[somecallmemike]

Hi,

I am very impressed with your offering to the virtual market, but I was wondering what your timeline for Proxmox VE 2.0 looks like? We are evaluating a variety of virtual products, clustering software, and SSI implementations in the hopes of building some redundancy and failover into our services and I really like what Proxmox can do, but I do agree with the review given here that it needs some more features to really stand out in the data center.

Thanks,

Somecallmemike

 

[tom]

Hi,

I am very impressed with your offering to the virtual market, but I was wondering what your timeline for Proxmox VE 2.0 looks like? We are evaluating a variety of virtual products, clustering software, and SSI implementations in the hopes of building some redundancy and failover into our services and I really like what Proxmox can do, but I do agree with the review given here that it needs some more features to really stand out in the data center.

Thanks,

Somecallmemike

things are moving fast and we will release features step by after finishing. I know a lot of people asking for a schedule but there is no fixed time line.

the new cluster concept will be based on the latest and greatest open source technologies and you can expect a lot - but its still not finished. Just watch out for news on our project pages and the forum announcements.

In the meantime you can think of using drbd (8.3.1 is included in our kernel) - but keep in mind there is a lot of manual work and the upgrade path back to a supported Proxmox VE server is probably not that easy.

 

[somecallmemike]

Tom,

Thanks for the quick reply! I look forward to your developments and the release of 2.0.

Mike

 

[Jorem]

I like the DRBD option for V2.0 a lot. But I need to set up a server in the next week.

Is it possible to update a running system later and add the DRBD to sync (network RAID1) the servers?
Then I can set up the server now and use rsync to sync the backups for now and move to DRBD with V2.0.

 

[tom]

I like the DRBD option for V2.0 a lot. But I need to set up a server in the next week.

Is it possible to update a running system later and add the DRBD to sync (network RAID1) the servers?
Then I can set up the server now and use rsync to sync the backups for now and move to DRBD with V2.0.

DRBD 8.3.1 Kernel module is already included in the current V1.2 - but setting up and configuring needs some expert advice. maybe you think of asking the commercial company behind DRBD for help (www.linbit.com) or other DRBD experienced people.

migrating a custom system back to a standard setup later will include some manual work and will not be possible without some manual adaption - worst case - reinstall. As 2.0 is not finished yet I cannot give details about what is possible and what not in kind of migrations from non-standard setups.

 

[mikeborschow]

Just curious...???...

Is the biggest holdup on V2 centered around creating smooth and trouble-free 'upgrade' capabilities? I for one will be happy to go to the trouble of having to move each VM one by one onto a new platform if that brings it sooner.

Put iSCSI or even NFS for shared storage into the mix and PVE will see impressive growth in market share.

I am about to install ConVirt with great reluctance ... I would much prefer to keep PVE as my loyal virtual companion. Don't underestimate the loyalty base you have. Many of us are frustrated because we're being forced to look at other options.

Give us HA, shared storage and more raid options (including software with a huge 'use at your own risk' disclaimer) ... just let me enjoy that for 30 to 60 days and I, most likely along with many other loyal PVE users, will be to happy to pay for the features.

 

[fortechitsolutions]

Just a question to mikeborschow, as I follow through this thread. I'm puzzled why you might got with Convirt as a virtualization platform, rather than Citrix XenSource, assuming you had to deploy "now!" (ie, rather than wait for ProxVE 2.0 to arrive)

- high availability features maybe (?) can be setup with Convirt, and not supported on free Xenserver ?
- inherent desire to not support a 'crippleware loss leader' type platform?
- concerns over longevity of XenServer?
- ... ?

I guess I'm just not clear what would be seen as the 'key feature'(s) to pick convirt over XenServer, so I would be intrested to hear your thoughts on this.

My feeling in testing both products, was that Convirt appeared to be quite a bit more 'rough around the edges'; still usable clearly, but not as polished. But if there was a key feature one has that the other lacks, clearly that could be a major / deciding issue ...


Tim

 

[mikeborschow]

All good questions Tim.

KVM is primary target on the radar (where it should be).

Xen is out. No need to go through the laundry list here (you already aired some of it anyway). My attempt at Convirt is all KVM, not XEN.

While this may sound like a contradiction to what I said earlier, we've all heard the phrase; 'less is more' ... and there's the beauty in PVE. It's just a hair on the 'not quite enough' side for 'committee' thinking, which demands shared storage. Yet, the 2.0 roadmap clearly addresses the bulk of the demands. Deploying PVE 2.0 (production or beta) will be a MAJOR move to help get people to take off their comfortable 'old shoes' so they can try on some PVEs with their KVM laces tied up.

I agree with the arguments that battery backup is essential for running a raid if you are going to work according to best practices, especially in a production environment. Show me a shop where every technical implementation follows best practices..?? Software or hardware raid with no bbu, users should be able to determine the risks they are willing to take. There is always the option to force write-thru if you must step out without bbu, but to me, raid 1, even if it's software raid ... that's a no brainer. Unfortunately, I don't have my arms wrapped around DRBD, but a simple raid 1 is huggable.

Measuring risks; for me, running PVE with a single drive is a much greater risk compared to running with software raid 1 which leaves me with a real recovery plan in the face of disaster.

In a tapped economic climate, those that are already sold on virtualization (and using it) will not do any ‘wholesale’ switching over to PVE (KVM/QEMU/KQEMU) … but they are more likely to ‘dabble’ a try with it if they can do so at first without making more hardware investments over and beyond the obvious VT/AMD-V based machines they will need to use. Carving out a little LUN or loading some 'software raid 1' makes a huge difference between dev and prod for those willing to take PVE to the next level. And taking it to the next level on a greater scale is what drives momentum.

I've been using PVE since 0.9 in my personal little hosting business, but even there, the prospective customers that are willing to spend more are asking for storage ops that are not part of PVE core. You can’t tell [propeller-head] customers what they want – it’s the other way around. Yes you can share your reasoning, but ultimately, if they don’t buy your pitch, they won’t buy your service either.

Back to the larger enterprise – at the office, I don't have to deploy 'now' but I have to keep things moving. I unfortunately read a comment in a blog response that said V2 was slated for Q1-2009 delivery (https://www.montanalinux.org/proxmox-ve-review.html) and I thought I saw it somewhere else as well. So go figure, I started pitching PVE with the 2.0 features ... my bad, but still, I know that other people are equally as anxious as I am.

Talking about roadmaps; I like the road I'm on ... I am eager to get more people to take this path as well. Which leads to my final cliché phrase for this response … the natives are restless.

I save my last word for the PVE team and those that are part of the community: Great work – thanks!