Please watch the video – it such fun!
If you been following my blog, you’ll know that I have been selling off my colocation gear, with a view to return to a homelab. As promised I want to write a couple of blogpost about the why and how – and just share my experiences generally about the new gear I’m using. In this first post I want to talk about how I came to find myself in a colocation and why I moved.
When I started out in VMware in 2003/4 the chance of getting VMware ESX to work on entry level kit was slim to nothing. It’s a matter of amusement to me that my first attempt to install ESX 2.0 was to Dell Optiplex PC. Needless to say I didn’t get very far! I went out a bought two Dell 1650’s pizza boxes. These bad boys had 2GBs of RAM and Pentium III processor. It’s a matter of personal pride that I think I might be the only person who ran ESX on Pentium III. You see I’ve always believed in doing a lot with little!
My lab environment grew – and in the end I bought some factory-sealed HP DL385s off ebay. I did the processor upgrades and memory upgrade myself. At this point the gear was in my basement. I’d power it on in the morning when the wife had gone to work, and power it off in the evening when she came home. It was just too noisy to have on all the time.
Then in around 2006/7 something very special happened. Both EMC and NetApp donated on a long-term lone two arrays each (2xNetApp2020 and 2xNS20). This volume of gear was never going to fit in my basement, and so I went to my first colo. It was dreadful place, but cheap. Why dreadful? Their idea of power was big long white gang power extenders gaffer-taped to open racks with no security. NICE. What did realise at the time was although my storage pals had been very generous – they also gave me a colo bill to go with the gear. Any money I made through writing books was wiped out by colo charges. But heck, I had great gear and it was empowering me to do stuff I hadn’t done before. So you can’t complain.
Then I left the colo. Why? I had two black outs at the colo which didn’t inspire confidence. But I was also at the end of writing the SRM 4.0 book. The colo was costing me an arm and leg. I left. Unracked the gear and stacked in the dining room (to the delight of my wife!). And invested in a homelab – 4xLenovo TS200 with an IOmega NAS (it was one of the 1st small NAS devices to be supported by VMware). On that gear I wrote the first VMware View 4.5 book/guide. Not to shabby. Although the homelab was a financial outlay – by leaving the colocation within 6 months the savings in fees had paid for the kit. Then, an opportunity to write the first ever book for the VMware Press came up with SRM 5.0. It was back to a new colocation – this time a much more professional operation. At the same time I got access to Dell Equallogics arrays. Things were getting a bit silly. I had more storage capacity than I had servers!
Next I joined VMware. I’ve been running the colo since I join the company, and despite the resources available to me internally – I still found having my own rig was very useful in my work. So, why did a leave the colo?
1. My old server gear was getting very long in the tooth, and so was some of the NetApp storage (the EMC gear went back after the SRM book was completed to free up rackspace for more servers – the HP/Lenovo together provided two clusters). It was time to refresh the gear or get out.
2. Cost. With those colo fees I could bought another house – albeit not in the South of England, but from the Midlands up.
3. I think we’ve reached a point with storage both physical (Synology, Dobro) and virtual (Celerra, NetApp, HP VSA). That performance on the storage layer is as good on cheap NAS devices as incredibly good – and for me, if I want to do anything fancy you can use virtual appliance. In fact I used the Lefthand VSA on cheap, dumb storage to write the first ever SRM 1.0 book. The difference now is – so many storage vendors support this now, whereas back in 2006/7 you were pretty limited. Of course, there are some vendors who don’t do this yet. Over the last couple of years I successfully nagged my friends at NetApp to make their emulator publicly available. I’m turning my sites on Dell Equallogics now. 😉
The History…
So its with this history in mind which shapes my “design” (its a bit grand to call it a design) that continues to shape my decisions now. Firstly, I’ve always looked to have as “production” like configuration in mind. Secondly, because I’m concerned about devices not working – or not working in the future – I’ve always acquired gear that’s support on the HCL. I like things to work out of the box, and excessive work installing VIBs, creating custom images – doesn’t float my boat. Thirdly, wonderful though nested ESX is, I’ve always preferred to go native and work directly on hardware – not because of performance (although that could be a factor) but because I want as close to the real world experience as I can get.
With that said I do think that this might be a last lab that does run natively on hardware. One of my ideas to extend the lifetime of my gear is go-nested when its no longer works with VMware ESX version 9. After all if your running on vSphere5.5 with VMware ESX 9.0 nested then you don’t any hardware compatibility to think of.
In the end I went for the HP ML350e series boxes together with the Synology NAS… In my next episode I’ll write about what they were like to use.