Shulou(Shulou.com)06/02 Report--

Bidding instances are idle computing capacity in the cloud, and it is one of the billing methods for cloud providers-the other two are on-demand instances and reserved instances (prepaid instances).

As far as the server itself is concerned, there is no difference between the three, the difference lies in the business model. On-demand instances are charged according to the duration of use, and users only need to pay when they use them. Reserved instances allow users to purchase in "month" or "year" units. Bidding instances allow users to save up to 90% of cloud computing costs by using server resources currently idle in the cloud.

Cloud providers hope to take advantage of these idle resources by pricing these idle servers in the cloud at very low prices.

What should you know about bidding instances (Spot Instance)?

01 bidding examples are very cheap

Compared with on-demand instances, bidding instances are usually only 10-20% of their price. Compared with reserved instances, bidding instances are usually only 30-60% of their price. Bidding examples offer a large number of possibilities to save cloud resource costs.

02 the bidding instance may be terminated at any time

The risk of using bidding instances is that cloud providers can terminate them with little warning (usually a few minutes' notice). If your application needs to ensure availability, consistency, or data consistency, then when you use bidding instances, you need special management configuration tools to ensure these. Cloud providers will not provide SLA for bidding instances, and your application needs to deal with these interruptions that may occur at any time.

Take AWS Spot Instance as an example

EC2 Spot instances are the excess computing power of AWS (usually a small number of on-demand servers). EC2 Spot instances span all areas, availability zones, instance types, and instance sizes of the AWS.

If you look for a specific instance type in a specific availability zone, the bidding instances may be very inconsistent. This is because some instance types in specific availability zones are sometimes fully utilized and are not considered bidding instances.

For most companies, this usually means that the use of bidding instances is limited to certain development environments and data processing efforts.

However, by taking advantage of multiple instance types and availability zones, bid instances are much less unstable, because in a suitable availability zone, you have a good chance of finding a corresponding server for an appropriate instance type. By expanding the scope of instance types and automatically fallback to the by-quantity server when no bidding server is available, and if recycling can be predicted in advance, there is time to trigger intelligent migration (all migrations are behind the load balancer), thus ensuring that the application is always stable and available, and even some very good professional DevOps teams manage the system by writing a lot of scripts. As a result, some key workloads can also be run on bidding instances.

Companies and teams that are not willing to spend a lot of time or take risks can use a professional bidding case management platform to provide them with professional services.

By tracking the historical data of public cloud vendors for a long time, the cloud scaling group has trained proprietary and reliable price and availability prediction algorithms. The cloud scaling group uses this algorithm to accurately predict outages and fluctuations, so that the cluster can be actively rebalanced to prevent outages from occurring.

Cloud scaling groups reliably take advantage of the excess capacity of all major public cloud providers, such as Spot Instance (AWS/ Aliyun / Tencent Cloud), Low-priorityVM (Azure) and PreemptibleVM (Google Cloud), while eliminating risks and complexity, providing simple orchestration and large-scale management.

Cloud scaling group predicts the behavior, capacity trend, pricing and outage rate of Spot instances. Whenever the risk of disruption is predicted, the cloud scaling group balances capacity 15 minutes in advance to ensure 100% availability. This means that the application will always run on the most cost-effective instance cluster-that is, running the best-priced and available Spot instances when available and on-demand when unavailable, and leveraging all the prepaid instances you may already have to achieve cloud computing cost optimization.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.