Entertainment gossip preview revealed that various platforms have been gearing up for launch, only to find that there are only “scraps of material”, there is no massive business demand, and computing resources are idle…
Is there an “elastic” way to use computing resources as efficiently as possible while maximizing their value? Of course! Shence data online elastic computing and elastic storage functions, so that enterprises do not need to increase hardware investment for the peak computing demand, save money and save effort to “win” business demand!
What is the function of elastic computing and elastic storage?
As companies based on the analysis of the use of computing resources exist obvious periodicity, set “, cure and stock, check, intellectual “ce data foundation platform for the integration of god through flexible allocation of resources, can be as the change of the actual business needs, increases or decreases computing power at any time, avoid enterprise computing resources waste caused by the adjustment of the business needs; Moreover, Shence borrows the concept of traditional database archiving to support the migration of the data that is not often used in the data warehouse to the low-cost storage system, which reduces the storage cost while ensuring that users have a better product use experience.
1. Elastic computation: tidal computation force scheduling
By constructing a common computing pool, the computing capacity can be dynamically allocated. If an enterprise needs to expand its computing capacity to support its services, the system applies for additional computing nodes from the public computing pool to form a new computing cluster with the existing nodes to match computing operations during peak hours. After the peak, additional compute nodes are reclaimed by the common computing pool to save computing resource costs.
The technical principles are as follows:
First, k8S is directly used to manage the elastic resources of the computing pool. In K8S, each executor node consists of svC-Executor and pod-Executor.
Pod-executor exposes ports through SVC-Executor. Statestored, Catalogd, and Coordinator in the original cluster communicate with pod-Executor through ports exposed by SVC-Executor.
The POd-executor in K8S communicates with Statestored, Catalogd, and Coordinator through the hosts of the original cluster configured by CoreDNS to implement dynamic deployment of compute nodes.
Schematic diagram of Shence elastic computing Technology
2. Elastic storage: Cold data is archived and can still be analyzed
For some infrequently used data, the frequency of data access is constantly reduced, and even only used in the historical comparative analysis of the occasional inventory, which is called “cold data”. However, as the enterprise spends more time on data analysis, the demand and pressure on cluster storage resources become heavier. The smart solution supports archiving (dumping) of “cold data” to cheaper storage systems, thus greatly saving enterprises’ storage costs.
It is worth noting that when “cold data” is archived, Shenze still supports the archived data readable, deleted, and analyzable. The storage where the archived data resides is part of the overall system storage. Archived data can be queried without unarchiving. However, in the current situation of the industry, most of the archived data can no longer be queried, which is the most important difference between Shence elastic storage function and traditional database archiving.
Second, to escort the promotion activities, let god’s “two clouds” fully open fire
Shence “two clouds” refers to Shence analysis cloud and Shence marketing cloud. At the time of promotion, an e-commerce operation team through Shence analysis cloud and Shence marketing cloud “full fire”, helping the perfect launch of the promotion, while relying on shence elastic computing and elastic storage functions, through the three-step method, escorting the promotion activities.
Step 1: Archive cold data before making a big push
To better prepare for the big push, the e-commerce operation team archives relevant data in advance and transfers infrequently used “cold data” to lower-cost storage systems without affecting queries. As this part of “cold data” can still be analyzed and operated in each major system of Shence, the e-commerce operation team is also more “bold” to transform the data that is not often used into “cold data”, to “make room” for the subsequent analysis data generated by massive business needs and reduce the storage cost of data.
The second step: in the promotion, adjust the computing ability
Since the e-commerce operation team carries out strategic operation for the whole life cycle of millions of users through Shence marketing cloud, and recently, it has continuously issued activity coupons for various user groups, so from the beginning of pre-sale, the surging user traffic has brought massive user behavior data and analysis needs. Starting at this point, the electricity business operations team members will enter a state of “high speed”, by god, their policy analysis cloud real-time insight users all kinds of state, by god, strategy marketing do joint strategies and actions of the user’s portrait of cloud, and the whole process into god’s policy introduced based on the data flow of enterprise operating framework SDAF, Through the four processes of Sense, Decision, Action and Feedback, the operation closed loop is created to improve users’ perfect experience in the promotion process.
Step 3: Wrap things up and avoid waste
In a tense and exciting activity to an end, the electricity business operations team members from a state of intense work gradually return to daily, members will be more focus on big for checking, and through the strategy analysis of the major functional components of the cloud insight into user requirements, and through the marketing strategy for the next phase of cloud strategy operation, due to the calculation and analysis are less demand, The elastic computing function enables the system to automatically reclaim part of the computing capacity, avoiding the waste of computing resources caused by traffic decrease.