Multi-tier (more than one server for my application, load balancer, web servers, database server...)
I would love to have the option of setting up a multi-server application for more targeted and cost effective scaling. For instance, Say I could create a Wordpress site with a dedicated DB server, a dedicated load balancer, and a dedicated Web server. As my site grows, I can scale the specific part that is needed. If my site is gets a lot of views (reads), but doesn't make many writes, scaling my existing web server (or adding another instance) may be what's needed. If I'm running a WooCommerce site and start pushing the limits of my db server, increasing the resources or implementing replication might make more sense. It would be really great to have that level of control when growing a site.
Vlad Slobodkin commented
Yes definitely needed
+10 for this. This solution shall really benefit a lot of ventures
I'm really interested in this.
I agree! Since you are working directly with some of the largest cloud service providers on the planet it would be nice to offer a managed Wordpress Cluster service consisting of a server acting as a load balancer in front of two or more servers running Wordpress with two or more database servers with failover capability. Having the ability to have multiple servers running from multiple data centres from multiple geolocations would ensure the ability to run without any downtime. Actually, Amazon AWS offers the ability to create a self-healing, highly available, and high-performance application/web auto-scaling hosting capability now. Their Aurora RDS (MySQL Compatible) database service is incredibly fast too.
Also, Vultr is working on their BGP capability.
Operating these services requires a wealth of technical knowledge and a team to support them due to their complexity. I and many other people running massive online businesses would benefit by having some of this complex management outsourced. To my disappointment Cloudways is not offering these services currently.
It is often beneficial for speed (page load times) if the DB server is not the same as the web server, would be nice if this type of setup was offered so we can, for instance, have a 512MB VPS for the DB and a 1GB instance for the web server.
+5 to adding an easy to use, easy to do ... option to have separate database servers, Load Balancer, high availability ...
Yes! to simplicity, but that doesn't mean you can't add complex features while making them simpler and easier to use for the "masses" :).
Take these complex aspects of hosting and make them into simple to configure, deploy & manage solutions - just like what you did so far with single VPS. Then you would have a better alternative to the do it yourself self-manage solutions out there. Clouways would open up using this advance architecture to everyone. And this is certainly matching perfectly the your current customer profile. That would be a crazy competitive advantage in the space.
Keep in mind that the customers that go VPN route are not the Wordpress.com build your own blog type or $1.99 shared hosting. Are more advanced and with higher requirements and expectations when it comes to hosting.
While you can boot with a simple solution many will get to a point where they learn about and will want these options. Now its up to you if you want to retain these customers or are fine to have a big hole in the bucket so to say.
When I joined recently, I was surprised this wasn't already a feature.
AdminCloudways (Admin, Cloudways) commented
We are considering how to better solve the multi-tier / auto-scaling / high-availability problem. As you know, our main focus is simplicity, so we need to consider it in this context too.
We have some ideas about it but need to develop them.
Patrick Healy commented
Yes, this would be really helpful for folks running API's that support multiple geographies when you need to commit to redundant servers.
I think adding load balancer to connect servers is really helpful, as I was testing using web load test and 2GB DO server doesn't hold any 100+ concurrent users. also when tested using 500, 1000, 3000.. the test was aborted because errors threshold was reached at the early of the test