AWS migration recommendation

tonmo

Well-known member
A few Sundays ago I was perusing this forum and came across someone talking about migrating to AWS... This has been a topic of general interest for me (I had been running on a dedicated server), so when I saw a post from @fly offering his qualifications and offer to PM for help, I reached out.

He didn't ask for this testimonial, but I offered to do it because the service he provides has been excellent. It's a partnership, of course... The winning combination is someone who knows the details of my current hosting situation (me) and someone who knows all the ins and outs of a) migrating web properties and b) AWS (@fly). I run my sites as a hobby, and so I have a "regular job" which keeps me from working on it during normal hours -- @fly is flexible, available, and supportive of your schedule. Even if you know some things about migrations and AWS, the reality is that having someone to compare notes with ("the reason you're getting that error is because you forgot to put a "." at the end of the DNS record) is pretty invaluable. I'm definitely novice so the service was highly valuable for me... if I tried to self-learn, it would have taken WAY longer and I probably would have given up.

Importantly, I am saving a LOT of money by having made the switch. @fly's services are reasonably priced... and when you're done, he offers up some tidy documentation and you're off and on your way. So, basically -- if you're thinking about moving to AWS, I highly recommend reaching out to @fly.
 
Importantly, I am saving a LOT of money by having made the switch.
This is the item I've been unable to reconcile when looking into AWS myself. Everytime I do the sums, even approx., I come up with AWS being more expensive.
Would appreciate if you could tell us more about your before and after costs, and how/where the savings were recognised?
 
it's a bit embarrassing... I feel foolish for overpaying for a dedicated server for so long (maybe I shouldn't say "overpaying"... but if nothing else, I was paying for more than what I need). Without being super-specific, consider this:
1) AWS offers micro instances free for the first year.
2) my monthly costs (after the first year) is estimated to be 28% of what I was paying for my dedicated solution. We'll see if that holds up, but currently I'm pretty confident it should at least say under 50%. That's a lot of savings.

There are plenty of articles out there on the benefits of AWS so I won't get into that (temporary scaling, pay as you go, etc.)... this all factors in. Suggest dropping @fly a note for further feedback! I have no skin in that game btw and don't otherwise know him. Worked for me, though :)
 
This is the item I've been unable to reconcile when looking into AWS myself. Everytime I do the sums, even approx., I come up with AWS being more expensive.
yeah the bandwidth costs are what kills AWS EC2 for me at US$90/TB. Traffic /bandwidth surge will be costly as someone on this forum reported before 2+ TB spike costing costing them >US$200 in bandwidth costs.

10TB bandwidth will cost you US$900 ! I could get a decked out dedicated server with 10TB per month included paid for 6-10 months for that AWS bandwidth price @10TB
 
Last edited:
Traffic /bandwidth surge will be costly as someone on this forum reported before 2+ TB spike costing costing them >US$200 in bandwidth costs.

True.... Note, AWS does offer billing alerts so that you can be informed and react accordingly in the event of such a spike.
 
True.... Note, AWS does offer billing alerts so that you can be informed and react accordingly in the event of such a spike.
yeah i use AWS services but only EC2 for testing. But use AWS Route53, S3, SES, SNS and Lambda in production.
 
I feel foolish for overpaying for a dedicated server for so long
Right, so maybe your usage and bandwidth didn't justify a dedicated and thus a right-sizing (it could have been a cheaper VPS, rather than AWS) gave you the realised cost savings?
So, if one was paying around $125 US p/mth for a dedicated (and usage justified it) with 1.3Tb per mth bandwidth (CDN traffic excluded), then a shift to AWS is unlikely to show any cost savings? Likely the opposite? This has been my cost calculator findings.
 
Right, so maybe your usage and bandwidth didn't justify a dedicated and thus a right-sizing (it could have been a cheaper VPS, rather than AWS) gave you the realised cost savings?
So, if one was paying around $125 US p/mth for a dedicated (and usage justified it) with 1.3Tb per mth bandwidth (CDN traffic excluded), then a shift to AWS is unlikely to show any cost savings? Likely the opposite? This has been my cost calculator findings.
While I'm a huge fan of AWS and I think it's a great fit for a lot of people, there certainly will be use cases where it might not make sense from a cost perspective. On the flip side, there is more to hosting than just cost. It's up to you if those other factors make up for any possible increase in costs.
 
This is just another case of making sure you find the right type of hosting for your site. All the hype is VPS/Cloud these days, and to an extent, dedicated servers. While they most certainly have their time and place, sometimes people don't need them.

If you are saving money with AWS, your site never should have been on a dedicated server in the first place. AWS is ungodly expensive for basically every resource (bandwidth of course being the worst). If you are truly using a dedicated server to its full potential, there is absolutely no possible way that AWS would save you money. Simply not going to happen.
 
AWS seems to be cost effective for tiny sites and mega businesses. A large forum would be expensive to run, and benefits are probably overkill. I doubt most forum is generating the revenue to make on demand scaling even necessary.
 
A couple more items for perspective on my own situation:

- have had my main site since 2000. Started on shared servers until that didn't work anymore. Moved to dedicated a long time ago (can't remember when). Not sure AWS was even a viable option at the time.

- I do indeed have reasons beyond cost for wanting to use AWS. As I said my sites are a hobby but it does aid my "regular job" to experiment in this space. It's obviously a popular platform and having an active aws account allows me to better understand how it works, including extended features which I may use in the future.

That said, I wouldn't have done it if I didn't think it would help me get back to the black through adsense / amazon cpm alone. If cost projections hold up, this is very financially viable for me, much more than what I had before.

My objective was never / has never been to find the cheapest host I could find. I've had some really bad host experiences in terms of reliability and support.

AWS can be daunting if you don't know what you're doing, which is where my recommendation comes in. I'm sure there are plenty of people out there feeling a bit stuck on a dedicated server; migrating to a potentially less expensive known entity like AWS is an attractive alternative (of course YMMV), and if you need some friendly and understanding help to get this done, @fly is your guy :)
 
Last edited:
If you are saving money with AWS, your site never should have been on a dedicated server in the first place.
+1 sizing the right server and specs for your needs is the key. That ultimately means understanding your web app(s) usage requirements and understanding your web app and your server/configuraion. This means regular monitoring is key. Some of the paid clients I consulted for saved heaps after I went through their setups from 7x web server clusters costing them US$10,000/month they went to 3 server servers handling 10x more traffic capacity and 65+ % faster page load speeds at US$2,000/month. AWS EC2 bandwidth wouldn't be an option at 40-60TB bandwidth consumed per day (60x90/TB = US$5,400 per day (US$162,000 per month) in bandwidth costs on AWS EC2)!

If you earn an income $$$ from your site/forum that can cover the cost of AWS EC2 bandwidth comfortably then yes maybe AWS EC2 and AWS cloud services would be beneficial i.e. ecommerce site with auto scaling requirements during peak shopping/sales seasons.

I'm sure there are plenty of people out there feeling a bit stuck on a dedicated server; migrating to a potentially less expensive known entity like AWS is an attractive alternative (of course YMMV)
probably better to hire someone or learn how to properly monitor and optimise your dedicated servers if you're stuck. This is a skill or knowledge that pays off for the long term - understanding your web app usage requirements and properly scaling as needed. It's a skill that you need even after moving to AWS cloud environment as not having such skill can be costly on AWS with autoscaling etc !

These days dedicated server specs are 10x times faster than Intel Xeons 6 or 7 generations old and those older generations were capable of handling 100,000s of users when configured and paired correctly with their server stack setups etc. These new modern Intel Xeons have more power than the average forum site needs when tuned, setup and configured correctly :)
 
Last edited:
I switched to AWS and its a steep learning curve. The first year AWS gifts you around 500-700 USD worth of servers (one server for your application, one server for your database, one server for memcache, one server for elasticsearch), as well as 50 GB for Cloudfront and many other incentives.

If you know AWS architecture, you can make full use of this huge gift and therefore reduce your costs significantly. You can also triple your server costs.

For me the beauty of AWS is to start as much servers as you like within a minute. AWS is actually great if you want to learn about servers, as you can work with a handful of them at the same time, you can always terminate them and if you are done with playing, terminate them all. With hourly billing, this is very cheap. You register a new www.domain.com ,add it to cloudfront, auto-generate signed https certificates, everything works within minutes.
 
Yeah learning on AWS is great and some things are cheaper or more reliable than doing it yourself like AWS SES and AWS Route53 for geodns/geo latency specific DNS (not regular DNS).
 
Yeah learning on AWS is great and some things are cheaper or more reliable than doing it yourself like AWS SES and AWS Route53 for geodns/geo latency specific DNS (not regular DNS).
And that's the 'core' of cloud services. It encompasses everything you can do yourself, but is packaged in such a way that its usually quicker, easier, and sometimes more cost effective to implement.
 
Top Bottom