WordPress Cloud Hosting Demands Special Attention to the Plan of the Architecture. Universal Optimized WordPress Cloud Hosting is a Pure Fuss. Cloud Server Instances are Multi-tenant, service continuity and scalability has differences. Readymade WordPress Cloud Hosting are Mostly Ponzu Schemes. Depending on your need and budget, you can choose between DigitalOcean, HP Cloud, Rackspace, Amazon and IBM as IaaS Service Providers. You can use OpenShift or Heroku as PaaS for building your infrastructure. HP Enterprise Cloud is the best solution so far we have seen and used, surpassing Rackspace far beyond. DigitalOcean is for cost saving. HP Cloud and Rackspace will be similar due to the underlying OpenStack Cloud Computing Software.
WordPress Cloud Hosting : Why Readymade Solution is Not Possible
Whether your website needs more computing power only during the publication of the posts or sudden increase of traffic spike can be expected – these are not possible to plan on those reseller web hosts named as WordPress Cloud Hosting.
Rackspace provides such service at very high cost per months (DevOps – $5K USD + Usage Charges). These Ponzi Companies usually use DigitalOcean or Linode to build an infrastructure. Never opt for vendor locked products. Some measures charge as per thousand visitors – it is an abnormal metrics in hosting. These idiots can not define the Protocols and Standards.
DigitalOcean charges $5 – $10 per month per instance for instances with lesser virtual memory and less number of virtual cores. If following any one’s free guide make you to fight a bit but can deliver an optimized setup, why you will use others’ overpriced product or Service? All – DigitalOcean, HP Cloud, Rackspace, Amazon or IBM has their own datacenter, Public Shares and many metrics which makes them white in real.
These ready to use over promoted “WordPress Cloud Hosting” are another face of HostGator Like Third Class Shared Web Hosts. If Intel starts direct IaaS service, obviously we will add it in the list. Do you ever use a Made in China non QC passed Toothbrush? Made in China has been a phrase to indicate bad quality. The literal meaning of Made in China is not bad – from Apple to many Chinese brands manufactures in China. If you do not put a Made in China non QC passed Toothbrush inside your mouth, why you will use a kind of Made in China non QC passed “WordPress Cloud Hosting” at the same cost?
WordPress Cloud Hosting : Example of Basic Scalable Architecture
For very low visitor website (less than 300 page loads per day), one 512 MB to 1 GB instance is more than enough. You can frankly use IP-based or Name-based vHost configuration to run such 5-6 domains on one server.
We are Providing an Example of Basic Scalable Architecture where the number of page loads is more than 2000 per day (count the number of your page loading inside WordPress Admin panel multiplied by 3, as they are never cached) and not more than 3000 per day. You need to adjust for higher need.
One Database Server – Minimum one 2GB Instance running MySQL or MariaDB, optimization tested with Major Hayden’s script. 4GB Instance server will give 2 virtual cores. 2 GB usually have 1 virtual core. You will use InnoDB as Engine. Database Server or Application Server is the main thing for a web software – all posts are saved there.
One FTP Server – Minimum one 4GB Instance running Nginx with either PHP5-FPM or better HHVM with PHP5-FPM fallback. Use one 2GB Instance if your budget is less. We have all guides – you can read our guide on Nginx Plus too. You will use XCache with WordPress. XCache directly suck the RAM, it is important to check the load average.
Highly Scalable DNS – Whatever DNS you use, should be highly scalable, offer more than 2 name servers and 50 years TTL. Rely us, only 50 years TTL can decrease the page loading speed.
Good Own CDN – Using others’ shared CDN is a bad idea. All big brands commonly uses Akamai as direct client. It depends on your budget, how you’ll use.
Memcached – Not required. XCache itself can push so fast with tweaked Nginx setup, Memcached will not be needed. Memcached will demand another server instance, it can increase server the response time, First Byte Loading Time etc. With 4GB Instance, you will not get much advantage with Memcached, XCache can not be used simultaneously for WordPress; so XCache is better. Redis is not dependable for WordPress.
Optional Front End Server – It is better to use a low power Nginx/Varnish server to face the traffic. The real IP of the server remains unknown to the hacker. It is not possible to try SSH by a hacker, if you are careful and instance’s IP is hidden.
Backup Server – a low latency server, preferably on different datacenter which will take backup after one cycle of full caching by W3TC plugin has been done. It will take backup of MySQL server and wp-content file.
OpenStack Server Snapshot – It is mandatory to take manual OpenStack Server Snapshots before any major work on SSH or update.
Total cost is not really high, than you are thinking if you use DigitalOcean. $60 / month will be the average cost. Do not rely on test data – 10000000 hits with $7 per month server – they are test situation, not real situation. Do not use Swap or Virtual Memory – this is not suggested.
W3TC like plugin will take a bit time to create the full cache after publication of a post. This time, after publication of post is the best time to check the server load. Obviously, if your visitor trend is increasing, you will tend to use more RAM, if decreasing, you will use less RAM. Never put load immediately after publication of post, scores will be a bit bad. naturally as the servers are working :
If we do a bit cheating like restarting the MySQL server, it will load faster for initial few minutes. How many months you can keep the MySQL server up and running with same performance, is the level of right tweaking.