Does your application require Consistent, Predictable, and Reliable Cloud Services?
By Caitlin Rice | Apr 09, 2019
Every Cloud Service Provider wants to deliver highly consistent, predictable and reliable services every day to their Customers. However, given Cloud providers are not immune to the challenges Enterprises have faced since the dawn of computing services (hardware lifecycle management, continuous software patches, network latency, system integration, etc.) services delivery can be inconsistent within and across like service offerings. Highly variable service is driving inefficiency through over provisioning (increased cost) or service degradation (service quality and performance delays) undermining the core Cloud value of simplicity, agility, utility.
Over the last 5 decades, the holy grail of enterprise computing has been to deliver consistent, predictable, reliable IT services for the Enterprise Business Operations. Let’s look at what this look like in the Cloud:
- Consistent Service Performance: How does your Cloud Provider deliver a service that meets your application needs at 10am as well as 2pm? How does your CSP deliver the same performance in US and in Australia and in Germany with the same number of like systems? How does your CSP Deliver the same service performance today, tomorrow and two years from now?
- Predictable Service Cost: Can you align your Application Transactions to Cloud Service Costs? Can your CSP deliver the same number of transactions today, tomorrow and 2 years from now for the same or lower cost per application transaction? Can you have a predictable cost model in Cloud that aligns with your business activity?
- Reliable over time: Can you get reliability Service Quality at a reliable Total Cost of Service to enable your business processes to rely on Cloud Services to delivery for your organization?
The move from on premises data centers to the cloud requires an even higher level of service delivery discipline to ensure that the migrations are successful, sustainable and cost effective. The Cloud Service Providers understand this and are striving to deliver CPR across their service offerings.
Krystallize Technologies recognized the need for a universal measuring platform to test, assess and quantify a Cloud Platform’s ability to meet a specific workload demand. Krystallize’s CloudQoS® generates the same workload on all systems and report out the ‘number of concurrent operations (transactions)’ that each platform can actually deliver. To demonstrate how important it is to test, track and measure performance, Krystallize started the CloudQoS® CSP Index in Mid-2016. Over the 3 years of CloudQoS Index information now exists and the data has shown that tracking and measuring Consistency, Predictability and Reliability (CPR) is critical to obtaining Enterprise Class Cloud Services.
Consistent Service: Based on this independent testing of Krystallize CloudQoS® – when looking across the overall Performance, Consistency, Predictability and Reliability, Google ranked number 1 overall (11 of 12 months in 2018). The ability for enterprise clients to see their services through the lens of Performance and Total Cost, CPR allows them to make confident business decisions on what provider will meet their application needs (migrate), which service is delivering for them today and tomorrow and will continue to meet their TCO objectives (CPR).
In the image below, when turning up 897 of the same servers in the same provider, same instance type, same configuration within US East availability zone – the Cloud customer received vastly different Service Capability Index measures (SCI) across the server instances provisioned that represented over 42% performance variation. Consistency in service performance is key in running effectively and efficiently in the cloud.
Predictable Service Cost: According to the Gartner Report (“Ten Moves to Lower Your AWS IaaS Costs”), “Through 2020, 80% of organizations will overshoot their cloud infrastructure as a service (IaaS) budgets, due to a lack of cloud cost optimization approaches”. How is it possible to manage costs effectively and determine a ‘Predictable Service Cost’? Understanding your cloud costs equates to more than just rack rate price of the cloud service provider, rather it is very dependent on the number of machines and their capability to deliver on the given application workload. If a single machine cannot deliver the performance an application requires then larger or more machines (more CPU’s, memory, or storage, network, etc.) are provisioned to ensure application delivery. When underlying supply of performance is not consistent it creates fluctuation in costs often leading to overspend. To truly understanding the application requirements (demand) with the real capabilities of the cloud (supply) is it possible to determine what your service costs are both now and in the future.
Based on the information collected by CloudQoS, when looking at CPR across the leading Cloud Service Providers in 2018, Google’s GCP ranked #1. (GCP lead the Consistency measures 9 months and ranked #2 in the other 3 months). Azure ranked 2nd, AWS ranked 3rd and IBM rounding out the top 4.
To bridge the gap between finding the ‘best fit, best cost’ service for a given workload, CloudQoS can capture an application workload demand and replicate it synthetically on target Cloud services to identify the best solution at the best total cost to deliver the application workload. This approach allows clients to more accurately price out what their post migration cloud costs will be on day 1 and through maintaining the performance, the cost on day 1000. Predictable cost comes hand in hand with consistent service performance.
Total Cost is made up of pricing hour time number of units required to execute the number of transactions per second the application required. In the chart below CloudQoS takes pricing information multiplies it by the number of units (fractional units) required to execute 100,000 operations a second (10,000 Web Pages per second). This example gives enterprises the ability to see and predict their actual service cost for their application needs. (Predictable Service Cost)
Reliable Service: Customers expect their service to be the same on day 1 as they do on day 1000. Service Reliability is more than just availability. Reliability includes the ability of underlying service to continue provide continuous, high quality, cost predictable service regardless of location, maintenance, patching, noisy neighbors, outages, upgrades, lifecycle management and other service changes outside of the enterprise customers hands. CloudQoS® continuously collects performance metrics and periodically benchmarks the service to validate services throughout the lifecycle. This enables the customer to ensure they are continuing to get what they pay for every day. In the chart below, the Service Capability Measure is the median measure across a sampling of 10’s or 100’s of like systems executing the same synthetic workload. However, the ‘range’ of performance is represented by the width of the bar showing how variable the service is across the estate over time.
In the 2018 CloudQoS CSP Index Google GCP ranked #1 in in overall Performance (Service Capability Measure). AWS and SoftLayer split ranking 2ndand 3rd with Azure ranking 4th.
CloudQoS® is pervading the performance transparency that the Cloud consumer requires to move and thrive in the Cloud.
How did your cloud perform in 2018? Did you get what you are paying for? Check out CloudQoS today to see where your provider ranked month over month.
For a full 2018 review, please contact Krystallize at firstname.lastname@example.org
Gartner reference: https://www.gartner.com/doc/3692517/moves-lower-aws-iaas-costs