If you’re a small company on a budget, building large-scale web apps can be tricky. In this short audio training session, Ryan Carson, director of Carson Systems, shares some tips on how to make your web app scalable, when you’re on a tight budget.
This session is from the 1-day event The Future of Web Apps, hosted by Carson Workshops.
For those that are busy, skip. More music and intro than content. Bottom line: “buy as much as you need to launch only” and “build scalability into your app”
Jon, sounds like you got something useful out of it ;-)
Some good tips here Ryan - thanks.
What are people’s thoughts on launching an app on a virtual dedicated server to test the market before committing to a dedicated server? This is something I’m weighing up myself at the moment as the savings would make it worthwhile, but I’d be keen to hear any thoughts.
Hey Rob,
I’d definitely go for a dedicated box. I think you’ll find that the price difference isn’t huge ($50 - $100). The last thing you need is someone else taking down your shared server!
Also, you’ll probably need to fine-tune the server a bit, which you won’t be able to do on a shared box.
Just my two cents.
Thanks for the feedback Ryan - it’s likely I’ll end up taking your advice. There are a few hosting providers doing a package under Linux using Virtuozzo. This is different from regular shared hosting in that it effectively splits a dedicated box into several virtual dedicated boxes - each can be rebooted independently and have custom software installed. You just share physical resources (ram, cpu and hd space) which each get partitioned effectively. If one virtual area crashes, it doesn’t affect the others, allegedly.
It seems like a good way of getting a foot on the ladder without risking significant investment, though I can see what you mean about the “what if” fear-factor of having to share a machine. I’ll have to recount the pennies…
I totally disagree with you Ryan. A virtual server can be an excellent starting point, and can be had for $15 bucks. A XEN server is functionally immaterially different from a dedicated server, and if you don’t have to spend the money don’t.
Just be aware that “Virtual Server” means a lot of different things to different people. There are many platforms. Commercial Virtuozzo, OpenVZ, XEN, flavors of VMWare, etc. Also, whether you opt to slap some kind of control panel (don’t - very few can go multi-tier) in there makes a huge difference.
Virtuozzo does NOT cleanly slice a server - I should know, I own a business that is an “official SW-Soft partner” and have been running it for years. The main problem with Virtuozzo has to do with updates and update services. Virtuozzo is difficult to patch, and you don’t use tools like yum or apt-get to do it. You have to centrally manage it, and the vast majority of companies rely on SW-Soft, who “patches when vendors release patches”, a strategy that often leaves them months behind current technology (or more).
Educating yourself on Virtualization technologies, how they compare line for line, and how you can scale out of them as your business grows (this is a major issue - most small business owners who start on VE/VPS/VDS don’t think about it enough) is actually critical to whether you should use them or not.
Also, if you’re running your online business on a $50-100 a month dedicated server you might as well fold up shop - those are not quality providers. If there’s one addum that’s true around deployment, you definitely do get what you pay for.
Come on guys, the best of all is start with shared server and after if things going ok then go for dedicated, you will save some money and time also.
I would suggest starting cheap and doing lots of load testing prior to going public. Then you have a pretty good idea how your system will perform for a given number of users. There is a useful review of some load/performance testing tools at http://wiki.rubyonrails.com/rails/pages/HowToStressOrLoadTest
(the article is valid for any web technology - it isn’t Rails specific)
You might find that a bit of performance tuning can save you some real money on hosting costs.
I am not a web developer or programmer, but have lots of experience with VMware and totally recommend it. We use the enterprise version at work (ESX) and use the free VMware Server for testing. VMware is perfect for testing and extremely scalable. We use two dual core processor boxes (ie 4 cores) with 8GB RAM and only 2 x 36GB mirrored local hard disks. The servers are attached to a SAN for the disk used by the virtual servers. This means we can use Vmotion to rebuild servers on the fly is anything fails or we need to allocate more processor etc. We have about 20 virtual servers per ESX host.
I recently started a few websites on the side, which are designed by others, but I got the hardware and hosting together. The bootstrap method of the above is to have one box with loads of local disk running Centos (Red Hat Enterprise Linux clone) and VMware Server. We lose out on the extra resilience and ease of use of Vmotion and a SAN but we can still migrate virtual servers manually and currently have 5 virtual servers running very well on one physical server. The physical server is totally redundant ie two of everything and RAID disks. I colocate the server in London for about £1200 per year. I bought 1 year old IBM hardware on Ebay for about £1000 and paid nothing for software as it is all Open Source or free. The best thing is we have a clear upgrade path to the ESX Server and shared disk when the sites are paying their way and we can afford “Enterprise” licensing costs…
Forgot to say - great site and see you at FOWA in Feb 2007!
Why you don’t post a text version? I’am foreign an it simple for for me. How about accessible? =)
Come on guys, the best of all is start with shared server and after if things going ok then go for dedicated, you will save some money and time also.
And what about Amazon EC2 and S3?
http://www.amazon.com/gp/browse.html?node=3435361
soo very very goodd thankss
That’s great! I’m looking forward to it
tskler
I Think İt is Very good information ….
Thanks…
I think you did a great job of pointing out the major strengths and weaknesses of the book.
making money…
Prof. ……