if it is ok to share it, how much you pay for servers? how many requests/seconds you got? how much memory you use? or what GC or AWS instances you are using?
Running co-located servers for email & web (marketing services). We amortise the cost to $3k a month ($1k hardware, $2k colo/transit) over 5 years. Total CAPEX was $65k AUD 5 years ago including the warrenty.
That gives us 3 Nodes with (192GB of Ram, 2x Hex core Xeons) and we have 7TB of Primary Storage w/ 12TB of secondary Storage. Due to the age the storage is all hard disk, but if we were to buy today it would be SSD Cached or even pure SSD for primary storage. VMWare manages the cluster and we currently have ~ 50 vm guests.
As we have our own IP space, we have over 1500 IP address available.
Connectivity is slow as it is hosted in Brisbane, Australia, but we get 500Mbit of connectivity which we can saturate easily without taxing the hardware. Cost and business need is the only things keeping this from being higher.
Web activity is typically either highly dynamic but low frequency (custom web apps) or static, but high volume (image hosting for marketing emails). Since the payloads are high compared to most pages / APIs I really don't have numbers to compare as far as requests/second. We also tend to not be running the latest in the web stack, simply to keep maintenance down (we tend to use what ships with the base OS (Centos 7) and then add one or two packages if needed (PHP7)).
Have you considered Softlayer/IBM Bluemix? There are Softlayer datacenters in Australia with great bandwidth to the POPs. By default your servers get 1G bandwidth and you can upgrade to 10G and up. You can bring your own subnets of course. PM me if you are interested in any details. Disclosure: softlayer engineer
I run my personal vps' and a site for a family member[1] on two "droplets" on digital ocean for $5 a pop and couldn't be happier with their service. Also their collection of articles is fantastic for finding out how to get different things set up or whatnot[2]. Love 'em! <3
Got a bunch of servers with Linode: personal GitLab, self-hosted Zimbra, Pgsql database, personal/family/fun x2 (python, some PHP, some go behind nginx), couldn't be happier, have been paying $120-$160/m for almost 3 years now. Now also behind free cloud flare account getting free SSL and http2.
1x$70 at OVH, 3x10GBP at Feral, 2x$5 + $20 at Vultr. OVH is mostly used for transcoding, Feral for storage and Vultr for my own blog and misc stuff as well as an active job I'm working on.
HA is High Availability. So that if one of the servers fails, you still have a server that can handle requests. Scaleway have a quick overview of how this can be done using their infrastructure: https://www.scaleway.com/docs/how-to-configure-round-robin-d...
$10. I used to have a coloed server, where I would run multiple virtuals with KVM, but found myself hardly using the instances. I migrated my core services to Vultr, and haven't had any issues.
I use an m3.xlarge AWS Elastic Compute instance to run a LAMP server for hosting websites.
On this instance I have 24 WordPress based websites, all for small businesses, so no major traffic hogs (I've not exceeded 1TB of traffic per month in 3 years) but I've set them all a 2GB storage limit. I've not yet exceeded 1TB of Magnetic storage.
No email being hosted on the server.
I spread the sites out over 5 IP addresses.
I've got it storing snapshots of the whole thing to a storage bucket too.
On average it's costing me just over $330 per month.
I upgraded to the m3.xlarge instance last year as previous, less powerful instances didn't have the juice to put up with some of the shocking WP plugins/themes that get put on it.
wow, Zoho looks really nice. I have a shared provider I main keep for their email. Their prices use to be $3 a month. Now they have moved them up to $15 a month.
31 comments
[ 3.6 ms ] story [ 75.8 ms ] threadAWS and Softlayer
That gives us 3 Nodes with (192GB of Ram, 2x Hex core Xeons) and we have 7TB of Primary Storage w/ 12TB of secondary Storage. Due to the age the storage is all hard disk, but if we were to buy today it would be SSD Cached or even pure SSD for primary storage. VMWare manages the cluster and we currently have ~ 50 vm guests.
As we have our own IP space, we have over 1500 IP address available.
Connectivity is slow as it is hosted in Brisbane, Australia, but we get 500Mbit of connectivity which we can saturate easily without taxing the hardware. Cost and business need is the only things keeping this from being higher.
Web activity is typically either highly dynamic but low frequency (custom web apps) or static, but high volume (image hosting for marketing emails). Since the payloads are high compared to most pages / APIs I really don't have numbers to compare as far as requests/second. We also tend to not be running the latest in the web stack, simply to keep maintenance down (we tend to use what ships with the base OS (Centos 7) and then add one or two packages if needed (PHP7)).
[1] http://amagiccarwashforsale.com/ (if it's hugged to death let me know
[2] https://www.digitalocean.com/community/tutorials
* ~1k request/second
* https://www.scaleway.com
* 8 + 8 + 2 = 18GB RAM
Tips: Use cheap hosting and build HA into it, server die all the time.
On this instance I have 24 WordPress based websites, all for small businesses, so no major traffic hogs (I've not exceeded 1TB of traffic per month in 3 years) but I've set them all a 2GB storage limit. I've not yet exceeded 1TB of Magnetic storage.
No email being hosted on the server.
I spread the sites out over 5 IP addresses.
I've got it storing snapshots of the whole thing to a storage bucket too.
On average it's costing me just over $330 per month.
I upgraded to the m3.xlarge instance last year as previous, less powerful instances didn't have the juice to put up with some of the shocking WP plugins/themes that get put on it.
To be honest, Zoho's service is pretty fantastic for free email.