New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Redundant Server for a mission critical website
I am planning to host a mission critical website that provide real time hotel booking services, which require 100% up-time and should have fast connectivity from different regions. Needless to mention the Redundant factor.
I was told by my Software Developer to consider MS Azure or Amazon AWS. I believe there are better smaller companies that provide personalized Managed Server that are equally better to these giant companies?
What do you propose?
Comments
We can provide redundant services within different locations in Frankfurt. Redundancy is done by utilizing typical ha configurations like mysql master+master, glusterfs and / or virtual machines which are running in a cluster configuration.
If Moldova it's OK for you, we can do this.
Stating a budget.
get 2 servers from different dc/network, use HA
This seems to happen a lot more. A software developer (who has more training and experience) who built the software (they know exactly what's needed) recommends to sysops or the managers, solutions, and those said individuals believe that they know better.
Nothing against you, just venting - although, Azure and AWS are designed for critical workloads in mind (geo-replication, 100% up time, insight tools, etc.).
Redundancy must be achieved on the application level, not on hardware / VM level. If your website says "Error establishing a connection to database" on every page, then you should consider the system as non-operational, yet the "VM" is still online.
What I do is I have a primary engine, a backup one, and a 3rd verification engine, each in a separate datacenter. The backup periodically makes requests to primary engine. This request does a checkup to make sure everything is running smoothly and the expected result is returned. If this fails 3 times, the backup engine will call the 3rd server to verify that the primary engine is down. If the backup & 3rd verification engine agree the primary engine is not working as intended, an API call to cloudflare to change the primary A record(s) of the website occurs.
It's good to use something with high uptime but don't put all your eggs in one basket. Even AWS goes down and when it does, you need to have a solution in place.
Hello,
Sent you a PM.
I've managed Clusters in Amazon, 500, 700 servers using the latest technology to run HA websites.
Sounds interesting, How are you handling checkup requests?
There is no reason you shouldn't aim for redundancy at all levels.
You could get a VM/Cloud Server which is inherently resilient to hardware failure. It also makes it easier to migrate or upgrade as needed, so unless you really need the power, it may be preferable over dedicated servers.
You should then consider 2 separate locations in case of network issues in 1 location.
Then your application would need to be able to keep each other up to date (maybe working on a master/slave // primary/secondary basis.
I don't imagine a true 100% uptime solution being cheap.
With all that in mind, we can offer you an HK location. We have redundant everything!
But this doesn't protect against software/OS errors. As mentioned above, for true 100% uptime, you would probably need to build redundancy and high availability directly into your application.
It calls a custom script on the primary engine which that tries to pull data from the db, compute some stuff and output a result. This result shouldn't change so if it does, something is wrong.
Correct. I'm not saying don't care about redundancy at the hardware level, I'm just saying don't stop at the hardware / VM level. It needs to be all it way up the stack.
1000% agree. I see many people ping an IP and assume its online when it responds. If its a website, then have a file that computes an output and check against it. EG, a php file that connects to a DB and pulls a record, then does some maths and outputs the result. That result is always going to be the same so if its not then... Issues.
Hello,
We provide a fully redonded Cloud service that can be used to deploy vms in France, Germany, Netherlands, Singapore and soon Switzerland.
By defaut VMs are protected by a sw firewall and you get fast r/w thanks to SSD drives.
API access is also provided to help customers to automate some tasks.
Whatever you do, do not consider Azure for anything mission critical. It is the cloud service with most downtime.
Amazon EC2, by itself, won't give you 100% uptime. One of the EC2 we are using just randomly rebooted weeks ago, and you can imagine it's due to some hardware problem and they have to do dead-migration for it.
With Amazon or Google or Azure or w/e, you have to utilize your own solution to do high availability. Distributed file system, database replication, CDN, anycast, etc, should be considered. Amazon provides multiple locations and multiple availability zones, and that helps to create a distributed system.
Whatever...we use it at ${DAY_JOB} and it's fine.
Netflix (heavy AWS user) has a script that randomly reboots servers. At any time, any VM could be rebooted with no notice. This approach produces a very fault-tolerant design.
I didn't know that. I must admit, I like the idea, its pretty smart.
But yeah, as other mentioned already for OP, no system can be 100% fault proof, it's just impossible. You may only get a 100% or even 2000% SLA, for a lot of money, but SLA doesn't mean it won't break. You will need to invest in a redundant solution. Even the biggest sites on the internet, that spent millions on redundancy and 100% uptime, did have and will have some issues where users will be affected.
Depending on the project, setup a small cluster in multiple data centers, and you should get close to your 99.999% if its properly done.
Of course it can be fine, depending what you do and how you do it.
Btw i was wrong, google's cloud had more outages than Azure last year
>
source - http://www.networkworld.com/article/3020235/cloud-computing/and-the-cloud-provider-with-the-best-uptime-in-2015-is.html
Use multiple providers in different parts of the world.
One server does not guarantee 100% uptime.
kms hosting's (post 1)multiple frankfurt solution sounds good considering its peering status.
In requests like this, quotes ought to be in the form of a range, because so many things are variable, including how distributed-friendly your app stack components are.
Sharing relevant details of your stack and most important geographies, will help solutions and quotes.
Yeah, we have the capabilities to provide redundancy through bgp anycast, do ddos protection (which is often important for booking sites), setup layer2 interconnections beetwen these colo's to directly synchronize the corresponding files / databases via 1GE / 10GE / multiple 10GE transport - so, I would say everything what is required to make such project highly available, is possible with us
We also have a great experience with GlusterFS, MySQL Replication and HA Loadbalancer Setups in HA Environments. Regarding loadbalancers, we can loadbalance the traffic on layer3 before the requests hit the loadbalancers which then do layer7 balancing for multiple webservers to gain more and more redundancy
Hello. For you needs it is enough to get 1 server and to load static content via CDN - this can solve all your problems.
This is a correct point. Can offer you managed hosting; basically 99.99999% yearly uptime and i'm ready to compensate any downtime which should not ever happen. Pm me if you want to give it a try.
Are you looking for redundancy at under $7 a month?
And you're asking on LET?
not sure , why I am getting notification of this post , every time , when some one replies it
If you need a UK based managed service with 100% uptime, contact us via live chat at http://hostingspecialists.co.uk. We offer a few of these services already and the companies using us are very satisfied.
Perhaps you starred it...