has anyone ever hosted a high traffic website with these two companies? I would like to know who has the best service and servers. Thanks
I haven't tried Softlayer, the service offered by Rackspace is pretty awesome though, I've only had a couple of issues over the last two years and they resolved both within minutes.
One issue in particular was caused by multiple disk failures which resulted in one of my instances going down, they recovered from this with no data loss within about 15 minutes, which is fairly impressive.
I've had a server at rackspace, they did what i expected them to do. They have an awesome support though. I don't think any one could beat that.
I had a site which got around 60k uv daily and it worked nicely. I do not remember any serious issue on their part, unless it was a bug in my code.
As for softlayer, I've never had them.
I just had a downtime of 4 hours with SoftLayer's CloudLayer service and then 6 hours the next day because of host failure. The first time they just rebooted the host, without migrating our instance. The second time, they were really slow to move the instance to a new host, poor at communicating and favored investigating their physical host rather than making sure my virtual machine was up and serving 1000 clients.
This means nothing ...
"High Traffic Web Site"
Not trying to be harsh but want to re-focus you attention from traffic numbers such as page views or visitors per month to more operationally relevant metrics, e.g.
- HTTP Request/sec
- SQL queries/sec
The above metrics are they keys to picking underlying hosting infrastructure. A couple of cases:
Case 1: 80 Million page views/mo, Entry Level Hardware
I have one client with that does nearly 50TB and bandwidth and over 80 Million page views/mo the last time I checked. He uses a mid level dedicated server and two cloud instances for static content delivery. The cloud instances are tiny, single CPU low ram instances specifically tuned for their task.
Case 2: 10 Million page views/mo, 6 Server Cluster
For this client, we have 2 high end dedicated servers and 4 mid-tier cloud units. The high end units are $1000/mo systems. Dual Quad CPUS, tons of RAM, and Solid State RAID 10. They still have capacity problems.
The first client's operations are super clean and efficient. Minimal database queries, aggressive caching, and other tricks to keep the need for resources low.
The second client has a legacy application with know SQL query and other issues. Due to lack of developers, they have to over buy on the hardware side.
What's the point?
Unless you have profiled your application AND benchmarked the hosting provider's solution for your load, any advice is often just guesswork.
This being said, the biggest issue I've seen with all cloud offerings: AWS, SoftLayer, Rackspace, etc. is that Disk IO is very poor or unstable. You will see disk IO spikes unrelated to your own resource utilization. Somebody else is hammering the storage layer and you get hit.
This is why I still favor dedicated hardware if Disk IO is important to your application. Keep in mind that database write activity is can create higher disk IO. So if your app writes a lot of data, then be careful of cloud solutions.