Amazon has added new instance types t2.micro, t2.small, t2.medium
New instances cheaper than the old and introduce the concept of CPU Credit-ov.
CPU Credit is the internal accumulating coin, which is equal to 1 minute 100% load.
The credit is accumulated at a speed dependent on the server type. Also, the server type depends on how far it will drop the maximum CPU load in the absence of credit — baseline. The minimum maximum. Minimum as in the absence of credit maximum load sags not instantly, but gradually for 15 minutes, to be able to compensate for the lack of resources in case of need.
Stores the loan up to 24 hours.
the
So — t2.medium is at a very good price, with 4GB of memory on Board, with the ability to work 24 minutes to hours, or 9.6 hours per day, ie work a full day at the maximum load. I think this is a good suggestion.
Accordingly, the micro instance is able to work 2.4 hours a day with maximally load, and small — 4.8 hours. Check your logs it's not even a little.
It should also be noted that 9.6 hours is based, downloaded 100 1vCPU or 2vCPU percent for 50 percent.
(baseline respectively 20% or 40 x2%x1)
2vCPU on medium, both at 100% load can work no more than 4.8 hours a day.
For comparison table. it should be noted that t1.micro,m1.small,m1.medium it are based on annual prepayment based high load, and the new instances at any time you can improve or on the contrary to take the cheaper.
the
Prices for old instances taken from ec2 instances marketplace today, recently, these prices were much higher, so in General the current generation of faster and cheaper if there is no - load and there is no need to own the disks of the instance, new server types only support EBS.
By the way, Amazon cancel the old server types, such as t1.micro, m1.small, m1.medium. They can still take on the EC2 marketplace, but in General, they are considered obsolete.
It is no secret that the t1 was with throttling th, in fact I suspect that hence the first letter in the name of the instance, but in his case, information about how it will decrease the maximum performance was not disclosed, but now everything is clear and even with cloudwatch metrics.
There is a separate metric for balance of the loans and their use. That coupled with the fact that there was no longer need to make an advance payment(upfront) for servers in this class gives you the opportunity to automatically build a scalable configuration, which actually follows the same slogan — "Pay only for use".
By the way, the vast majority of servers, access to the body which I have invested in the limit t2.medium instances, I think I wasn't alone and the commissioning of these servers is very, very good alternative prior to this dance with a tambourine around light/medium/heavy utilisation.
Actually not even an option but a clear winner, as for many, the calculation of "take on year" is often not flexible enough, since it is unknown what awaits. Now in this segment is considered it becomes much easier.
It remains to wait for what will be at least t3.x2large, for projects where the peaks loads are frequent, demanding but brief.
Oficiala page: docs.aws.amazon.com/AWSEC2/latest/UserGuide/t2-instances.html
Article based on information from habrahabr.ru
CPU Credit is the internal accumulating coin, which is equal to 1 minute 100% load.
The credit is accumulated at a speed dependent on the server type. Also, the server type depends on how far it will drop the maximum CPU load in the absence of credit — baseline. The minimum maximum. Minimum as in the absence of credit maximum load sags not instantly, but gradually for 15 minutes, to be able to compensate for the lack of resources in case of need.
Stores the loan up to 24 hours.
| instance Type | Credit hour | Baseline | Price | Price per month |
|---|---|---|---|---|
| t2.micro | 6 | 10% | $0.013 | $9.50 |
| t2.small | 12 | 20% | $0.026 | $19 |
| t2.medium | 24 | 40% | $0.052 | $38 |
So — t2.medium is at a very good price, with 4GB of memory on Board, with the ability to work 24 minutes to hours, or 9.6 hours per day, ie work a full day at the maximum load. I think this is a good suggestion.
Accordingly, the micro instance is able to work 2.4 hours a day with maximally load, and small — 4.8 hours. Check your logs it's not even a little.
It should also be noted that 9.6 hours is based, downloaded 100 1vCPU or 2vCPU percent for 50 percent.
(baseline respectively 20% or 40 x2%x1)
2vCPU on medium, both at 100% load can work no more than 4.8 hours a day.
Performance
For comparison table. it should be noted that t1.micro,m1.small,m1.medium it are based on annual prepayment based high load, and the new instances at any time you can improve or on the contrary to take the cheaper.
| instance Type | Effective price | RAM | vCPU |
|---|---|---|---|
| t1.micro | 0.012$ | 0.615 GB | 1 |
| t2.micro | 0.013$ | 1GB | 1 |
| m1.small | 0.024$ | 1.7 GB | 1 |
| t2.small | 0.026$ | 2GB | 1 |
| m1.medium | 0.048$ | 3.75 GB | 1 |
| t2.medium | 0.052$ | 4GB | 2 |
Prices for old instances taken from ec2 instances marketplace today, recently, these prices were much higher, so in General the current generation of faster and cheaper if there is no - load and there is no need to own the disks of the instance, new server types only support EBS.
By the way, Amazon cancel the old server types, such as t1.micro, m1.small, m1.medium. They can still take on the EC2 marketplace, but in General, they are considered obsolete.
Throttling
It is no secret that the t1 was with throttling th, in fact I suspect that hence the first letter in the name of the instance, but in his case, information about how it will decrease the maximum performance was not disclosed, but now everything is clear and even with cloudwatch metrics.
There is a separate metric for balance of the loans and their use. That coupled with the fact that there was no longer need to make an advance payment(upfront) for servers in this class gives you the opportunity to automatically build a scalable configuration, which actually follows the same slogan — "Pay only for use".
Profit
By the way, the vast majority of servers, access to the body which I have invested in the limit t2.medium instances, I think I wasn't alone and the commissioning of these servers is very, very good alternative prior to this dance with a tambourine around light/medium/heavy utilisation.
Actually not even an option but a clear winner, as for many, the calculation of "take on year" is often not flexible enough, since it is unknown what awaits. Now in this segment is considered it becomes much easier.
It remains to wait for what will be at least t3.x2large, for projects where the peaks loads are frequent, demanding but brief.
Oficiala page: docs.aws.amazon.com/AWSEC2/latest/UserGuide/t2-instances.html
