8 New Features Recently Announced by AWS That Help Reduce Spend

We’re always on the alert for new angles to save money on cloud services, which is often exciting and sometimes challenging, considering the pace at which providers introduce new services. AWS continues to announce new features on a daily basis and tucked among them are some very useful new features that will help customers reduce AWS spend. In the spirit of sharing our miserly insights, we are dedicating our next few blogs to analyzing these announcements and how customers can use new features to optimize their infrastructure and reduce AWS spend. Going forward, we will write similar analysis on a quarterly basis.  Following is a list of 8 features AWS announced during April, May 2018.

Reserved Instance (RI) Coverage using AWS Budgets Alerts

Announced: May 8, 2018

Managing RI coverage is a significant challenge for many customers with applications that are growing or changing. Prior to this announcement, customers needed to keep a constant eye on RI provisioning and coverage in order to maximize RI coverage and minimize spend.

This new feature allows customers to set up RI coverage thresholds so that alerts will be sent when coverage falls below a specified threshold.

Cost Reduction Opportunity

The benefits are obvious for customers that take advantage of RI pricing models. Being constantly aware of the current utilization and RI coverage enables customers to minimize wasted spend.

Note:  Despite its huge popularity, FittedCloud is not a big fan of the RI pricing model. We believe applications are more dynamic than customers realize, and the pricing model places a huge burden on customers to manage coverage. Many customers are learning it requires constant monitoring, purchasing and fine tuning. In our opinion, pricing models such as RI, Spot are simply not sustainable in the long run because of the additional management headaches.

Optimize CPUs for Amazon EC2 Instances

Announced: May 8, 2018

The Opti-Miser covered this feature in an earlier blog which can be accessed here.

Amazon EMR now supports M5 and C5 instances

Announced: May 7, 2018.

AWS Announced M5 and C5 instances in November of 2017.  M5 and C5 are the latest instance types in their respective EC2 families offering better price/performance.

Cost Reduction Opportunity

Customers running EMR on C4/M4 instances can significantly reduce costs by switching to C5/M5 instances. AWS claims that C5 instances offer a 25% price/performance improvement over C4 instances and M5 offers a 14% price/performance improvement over M4 instance types.  Strictly from a price reduction point of view (without considering improved performance), M5 instances offer 4% and C5 instances offer 9%-13% cost reduction over Linux on-demand pricing. An earlier Opti-Miser blog explains the cost benefits and caveats of switching to M5/C5 instance types.

Elasticsearch Service Offers Additional Cost Savings with Reserved Instances

Announced: May 7, 2018

This new feature allows customers to purchase RI for Elasticsearch. Customers that run Elasticsearch 24×7 can take advantage of this new pricing option to reduce cost.

Cost Reduction Opportunity

By purchasing RI with an all upfront 1 year term payment, customers can save 35% over On Demand pricing, while a 3 year term offers 52% savings. RI purchases with partial upfront 1 year term can save 33% and 3 year term savings are 50%. For no upfront, it’s 31% and 48% respectively.  A sample savings calculation on the M4 family in the Ohio region is shown below.

Amazon EC2 H1 Instances Price Reduction

Announced: May 4, 2018

This is a straightforward 15% price reduction for H1 instances. Price deductions are available in US East (N. Virginia), US East (Ohio), US West (Oregon) and Europe (Ireland).  Customers need not take any action to take advantage of this price reduction.

RDS Cost Savings Using Reserved Instance (RI) Purchase Recommendations

Announced: April 19, 2018

The Cost Explorer has been expanded to add RI coverage for RDS. Now the same RI analysis and coverage recommendations that AWS has offered for EC2 resources also includes RDS resources. As mentioned earlier, identifying RI purchase can be a significant challenge for  customers. This feature makes it easier to analyze RI utilization and coverage for RDS resources to ensure optimum RI purchases and minimize wasted spending.

Cost Reduction Opportunity

For customers that want to take advantage of RI pricing options, significant cost savings opportunities are available. While there are a plethora of RI pricing options, the following examples provide a general idea of possible cost savings.

All upfront 1 Year Term – 30%-47% (bigger the instance better the savings)
All upfront 3 year Term – 55%-65% (bigger the instance better the savings)

Amazon S3 Select Is Now Generally Available

Announced: April 4, 2018

S3 is one of the most popular services used by AWS customers. A typical application stores data on S3, uses compute instances to read and transform data, then stores the results back on S3 for further consumption. A typical process brings entire S3 objects into compute nodes where the data transformation is performed. For many customers this could be a significant amount of data. S3 Select allows applications to retrieve a subset of data that resides in S3 objects using simple SQL expressions. This results in significant application performance improvements and, in many cases, helps to simplify data processing code/algorithms. However, it does require data to be saved in JSON or CSV format.

Cost Reduction Opportunity

Applications that can take advantage of S3 Select can realize significant cost reduction primarily from lower data transfer costs. Let’s use an example to help explain.

S3 standard data size – 10TB
Data Transfer cost – $900/month (Entire data transferred, $0.09 per GB. Up to 9.999TB/Month)
Data transfer reduction using S3 Select – say 75%
Data Transfer cost – $21.75/month ( $1.75 for data returned by S3 select,  $20 for data scanned by S3 select (assuming entire data is scanned)
Savings on data transfer cost/month – 97.5%!!

S3 One Zone-Infrequent Access, a New Amazon S3 Storage Class

Announced: April 4, 2018

S3 One Zone IA is a low cost option for customers that do not require multi zone redundancy for infrequently accessed data. As you may know, S3 IA uses a minimum of three geographically separated availability zones to offer the highest level of resilience to AZ loss. But not every application requires multi AZ redundancy.  If you have data in this category, significant cost savings can be achieved by storing data on S3 OZ-IA.

Cost Reduction Opportunity

20% cost savings (compared to S3 standard IA) can be achieved by using S3 OZ-IA. It may be worth reviewing the S3 data to determine if there are candidates that can be stored in OZ-IA. As per Amazon, good candidates are secondary backup copies of on-premises data or data that is already replicated in another AWS Region for compliance or disaster recovery purposes.

Again, it should be emphasized that OZ-IA should be used only for data that are recreatable from other sources as there is no multi-AZ redundancy.

On the same day AWS also announced that File Gateway now supports S3 OZ-IA, offering potential cost reductions for those using File Gateway.

Other related blogs

About FittedCloud

FittedCloud is the industry’s leading public cloud resource optimization solution. It features machine learning algorithms that continuously analyze resource utilization and identify opportunities to reduce monthly recurring cloud infrastructure costs. Automated provisioning can adjust cloud resources according to load patterns, user-configured policies and other parameters. FittedCloud’s patented solution reduces costs up to 50% while eliminating complex manual provisioning processes and the risk of configuration errors. For more details please visit https://www.fittedcloud.com