WEB HOSTING COMPANY TO RECOMMEND!

[GPGT.org]

anyone interested in High Availability Cloud IAAS Hosting in Singapore Equinix DC? Cheaper than AWS and better than DO in terms of redundancies.

 

[jf]

anyone interested in High Availability Cloud IAAS Hosting in Singapore Equinix DC? Cheaper than AWS and better than DO in terms of redundancies.

hmm, tell me more.

 

[GPGT.org]

hmm, tell me more.

just checking for interest as I am not familiar with the demand in Asia region especially Singapore. Looks like everyone is using conventional shared hosting, VPS and dedicated. Those pros/veteran enough are already using Amazon AWS so its hard selling to them.

The cloud IAAS runs on OnApp and a typical instance of 2 core, 2gb ram and 30gb SSD cost US$16 (with HA).

 

[GPGT.org]

if I recall correctly (in b4 davidktw to correct me), an AWS instance itself do not have redundancies (HA) built-in. You need to setup HA or deploy another instance for fail-over setups which double the cost of the instance.

However a high availability (HA) cloud instance is self-healing and have redundancy built-in, so that in an event if your VPS instance failed for whatever reason... it is replicated and "respawn" to another node within the infrastructure without human intervention or downtimes. SLA is 100% for this kind of cloud setup.

 

[jf]

if I recall correctly (in b4 davidktw to correct me), an AWS instance itself do not have redundancies (HA) built-in. You need to setup HA or deploy another instance for fail-over setups which double the cost of the instance.

I can confirm that (except I wouldnt say that it exactly doubles your cost). That is the very nature of AWS, and they do make that clear (somewhere).

 

[davidktw]

if I recall correctly (in b4 davidktw to correct me), an AWS instance itself do not have redundancies (HA) built-in. You need to setup HA or deploy another instance for fail-over setups which double the cost of the instance.

However a high availability (HA) cloud instance is self-healing and have redundancy built-in, so that in an event if your VPS instance failed for whatever reason... it is replicated and "respawn" to another node within the infrastructure without human intervention or downtimes. SLA is 100% for this kind of cloud setup.

None of the highly redundant setup will ever offer you 100% SLA. That's an unachievable SLA, because data centre failures does happens and global and network power failures also happens. Failover take time and even DNS can take minutes in it's shortest across the global. This part I hope has clarify for all small or large setup.

One thing you are right is for AWS EC2 service, AWS did not sell any of the instances to be resilient to downtime. They go down for a lot of reasons and one of them is instance migration when underlying hardware becomes faulty.

Microsoft Azure is one of those cloud providers that sell that their instances are redundantly backed not only 2 copies, but up to 3 copies and even across the global deployment. I got this first hand information from my past engagement with Microsoft Azure, but it could have changed, I wouldn't know now.

Not that I don't buy in, but from a software architecture standpoint, I don't buy into such design because I know network latency, power availability, storage decoupling and all these play into the part of redundancies. If I have very elaborate setup 100 servers that is designed across the globals for the need to be globally redundant and also to serve high loads, I certainly don't require the cloud service provider to perform their redundancy design for me having that wouldn't come entirely free. Somehow the cost will be part of the hosting package.

If you are starting something small, like just a pair of servers for local high availability. Then I think it's okay since such small setup is normally within the same datacentre.

For AWS, the redundancies happens across data centre, so you will find such design by default not realistic since latency across the continent can be hundreds of milliseconds and if you rely on such underlying replication technology for systems like databases, which they themselves already have something as such, it's either double work or less control at some level. Remember the database definitely requires more replication/synchronisation details compared to underlying storage redundancies.

When it comes to cost, it will be necessary to measure the kind of service you are providing and its behaviour. If you are starting with just a pair of servers, the cost of AWS can be higher. But if you are doing it large, that's where AWS costing can go substantially low because you can use spot instances at a much lower cost, but requires a very different design where processing and tasks requires queue system or other forms of clustering techniques. When it comes to system backup, you perform full snapshots with incremental cost, these are also costs you need to consider.

With the kind of auto recovery redundancy, it is today possible to achieve in AWS using Autoscaling of min:1 and max:1 instance. Today AWS also offers a new feature from Jan 2015, which one can read more from https://aws.amazon.com/blogs/aws/new-auto-recovery-for-amazon-ec2/

 

[littleprincey]

if I recall correctly (in b4 davidktw to correct me), an AWS instance itself do not have redundancies (HA) built-in. You need to setup HA or deploy another instance for fail-over setups which double the cost of the instance.

However a high availability (HA) cloud instance is self-healing and have redundancy built-in, so that in an event if your VPS instance failed for whatever reason... it is replicated and "respawn" to another node within the infrastructure without human intervention or downtimes. SLA is 100% for this kind of cloud setup.

I think this looks more like marketing to me. High availability can only be achieved in 1 way and AWS has demonstrated this.

Multi-AZ deployment

Self-healing and redundancy which I labelled as "network raid" at best cases work the same as hardware raid and it goes down if the network fails. The most probable reasons for a server downtime is

1) Network downtime
2) OS corruption due to hacking
3) Hardware downtime

Hardware downtime is very hard to achieve these days with the newer server hardware and proper RAID 1/RAID 10.

 

[GPGT.org]

Self-healing and redundancy which I labelled as "network raid" at best cases work the same as hardware raid and it goes down if the network fails. The most probable reasons for a server downtime is

depends on how the entire network is designed and deployed, on the case of OnApp... redundancies and self-healing features will be useless if there is a serious datacenter/network outrage, upstream provider fault, natural disaster which cause power shortage/destroy DC or any unforseen circumstances that are uncontrollable by the DC itself. These scenarios are very very rare so thats why 100% SLA (with clauses) are given but if anything happens then you will be compensated according to the downtimes.

1) Network downtime
2) OS corruption due to hacking
3) Hardware downtime

Hardware downtime is very hard to achieve these days with the newer server hardware and proper RAID 1/RAID 10.

All the 3 factors listed are irrelevant to a true cloud deployment.

1) Network downtime - there are usually multiple network providers and upstream providers for network redundancy
2) OS corruption due to hacking - this is irrelevant and often due to users fault
3) Hardware downtime - a true cloud has redundant sets of hardwares on standby and in event of any failures, the affected nodes will be automatically migrated to another part of the cloud where everything are working fine.

if you are interested, read up at OnApp.com for more throughout explanations because I am still a layman on such tech.

 

[yh_lee]

What kinda traffic you bLock for hostwinds ? Http traffic ? I have totally no issue with it so far.. I only host my sites there.. Dun use for any other purposes or email...

I specifically block all traffic from hostwinds. After seeing a lot of persistent spam from whole blocks of ip addresses owned by them, that is what I'm doing with them.

 

[natnai]

None of the highly redundant setup will ever offer you 100% SLA. That's an unachievable SLA, because data centre failures does happens and global and network power failures also happens. Failover take time and even DNS can take minutes in it's shortest across the global. This part I hope has clarify for all small or large setup.

One thing you are right is for AWS EC2 service, AWS did not sell any of the instances to be resilient to downtime. They go down for a lot of reasons and one of them is instance migration when underlying hardware becomes faulty.

Microsoft Azure is one of those cloud providers that sell that their instances are redundantly backed not only 2 copies, but up to 3 copies and even across the global deployment. I got this first hand information from my past engagement with Microsoft Azure, but it could have changed, I wouldn't know now.

Not that I don't buy in, but from a software architecture standpoint, I don't buy into such design because I know network latency, power availability, storage decoupling and all these play into the part of redundancies. If I have very elaborate setup 100 servers that is designed across the globals for the need to be globally redundant and also to serve high loads, I certainly don't require the cloud service provider to perform their redundancy design for me having that wouldn't come entirely free. Somehow the cost will be part of the hosting package.

If you are starting something small, like just a pair of servers for local high availability. Then I think it's okay since such small setup is normally within the same datacentre.

For AWS, the redundancies happens across data centre, so you will find such design by default not realistic since latency across the continent can be hundreds of milliseconds and if you rely on such underlying replication technology for systems like databases, which they themselves already have something as such, it's either double work or less control at some level. Remember the database definitely requires more replication/synchronisation details compared to underlying storage redundancies.

When it comes to cost, it will be necessary to measure the kind of service you are providing and its behaviour. If you are starting with just a pair of servers, the cost of AWS can be higher. But if you are doing it large, that's where AWS costing can go substantially low because you can use spot instances at a much lower cost, but requires a very different design where processing and tasks requires queue system or other forms of clustering techniques. When it comes to system backup, you perform full snapshots with incremental cost, these are also costs you need to consider.

With the kind of auto recovery redundancy, it is today possible to achieve in AWS using Autoscaling of min:1 and max:1 instance. Today AWS also offers a new feature from Jan 2015, which one can read more from https://aws.amazon.com/blogs/aws/new-auto-recovery-for-amazon-ec2/

Good post. I been reading up on deploying Hadoop on AWS storage instances for my forex project. Some people think that's self defeating because of latench. What do you think?

Sent from Xiaomi MI 3W using GAGT

 

[davidktw]

Good post. I been reading up on deploying Hadoop on AWS storage instances for my forex project. Some people think that's self defeating because of latench. What do you think?

Sent from Xiaomi MI 3W using GAGT

I'm clearly not the expert in HPC. but if your going to do Hadoop in AWS, is there something you need for your own setup that AWS EMR cannot offer ? That's probably the first question I will ask

If you need really good performance or some form or assurance, you will want to use the following features available. Provisioned and optimised EBS or the local SSD stores, dedicated tenancy, instances with high(10gb) network performance, place these instances into placement group.

In the shared environment, you are certainly looking at spikes and performance issues, but if you consider the effort and cost to maintain your own HPC outside of AWS, your cost could be even higher unless you have 24x7 work to run and a ridiculously high CAPEX

 

[natnai]

I'm clearly not the expert in HPC. but if your going to do Hadoop in AWS, is there something you need for your own setup that AWS EMR cannot offer ? That's probably the first question I will ask

If you need really good performance or some form or assurance, you will want to use the following features available. Provisioned and optimised EBS or the local SSD stores, dedicated tenancy, instances with high(10gb) network performance, place these instances into placement group.

In the shared environment, you are certainly looking at spikes and performance issues, but if you consider the effort and cost to maintain your own HPC outside of AWS, your cost could be even higher unless you have 24x7 work to run and a ridiculously high CAPEX

Will have to consider. A lot more to learn about enterprise-level scalability. I probably will need to dig much deeper into the more theoretical aspects of database and server/network design to determine the optimal solution. For now the idea is to simply show a working prototype and proof of concept. I'm not worried about getting investors as long as I have a solid proof of concept, with scalability into and past the 5-year mark in place.

 

[davidktw]

Will have to consider. A lot more to learn about enterprise-level scalability. I probably will need to dig much deeper into the more theoretical aspects of database and server/network design to determine the optimal solution. For now the idea is to simply show a working prototype and proof of concept. I'm not worried about getting investors as long as I have a solid proof of concept, with scalability into and past the 5-year mark in place.

High availability design is about everywhere in the Internet, but what works or not highly depends on cost, availability of infra and choice of softwares and technologies.

I have done up creativity HA design in the past that exploit on the feature of the protocol, which is SyncML, thinking back it is quite RESTful in design consider it is back around year 2006.

Feel free to discuss if you want any ideas.

 

[attapchee]

Any good hosting for wordpress sites?

Advertisements on services only. no e commerce. less than 3 emails per domain.
Currently have 2 domains but looking to increase more.

Need some help here. Just ran from hostgator.

Still panting*

 

[Yes8888]

Any one using hostpapa is it good?

 

[honeypanda]

Sorry for popping up old topic.

Anyone tried exabytes.sg before? Or anyone has recommendation for company hosting with like 50+emails accounts.

 

[youhsueh]

Sorry for popping up old topic.

Anyone tried exabytes.sg before? Or anyone has recommendation for company hosting with like 50+emails accounts.

Hosted with Exabytes, server located in Malaysia, speed so-so only..

Also, I've tried Vodien, server located locally, speed is good of course.. But the price a little ex..

 

[honeypanda]

Hosted with Exabytes, server located in Malaysia, speed so-so only..

Also, I've tried Vodien, server located locally, speed is good of course.. But the price a little ex..

I tried vodien but they always have downtime. :/
Exabytes website say their dataccenter at singapore wor

 

[youhsueh]

I tried vodien but they always have downtime. :/
Exabytes website say their dataccenter at singapore wor

Times probably have changed.. In that case can give them a shot, go for monthly billing first..

You can also check out WHT "Shared Hosting Offers" or "Hosted Email Offers" forum. Might be able to get one which suits your budget + requirement.

 

[tonyc]

Just buy a Synology NAS and host at home... no downtime