Comparison between Alibaba Cloud Elasticsearch clusters and self-managed Elasticsearch clusters

Updated at:

Alibaba Cloud Elasticsearch is a fully managed Elasticsearch service that is 100% compatible with open source Elasticsearch, delivers targeted kernel performance optimizations, and includes the commercial X-Pack plug-in for free. It is ready to use out of the box and supports high availability, auto scaling, and pay-as-you-go billing. This topic compares Alibaba Cloud Elasticsearch with self-managed Elasticsearch clusters on Elastic Compute Service (ECS) across five dimensions: overall costs, cluster management, capability support, security, and availability.

Item

Alibaba Cloud Elasticsearch

Self-managed Elasticsearch on ECS

Resource cost

  • Alibaba Cloud Elasticsearch instance fee: for example, a cluster of three nodes with 2-core, 4 GB (sn1ne.large specification), each with 100 GB of storage, costs approximately CNY 13,000 per year.

  • Supports auto scaling, so you can flexibly change node specifications, node count, disk type, and disk capacity.

  • ECS instance fee: for example, three ECS instances with 2-core, 4 GB (ecs.sn1ne.large) and 100 GB data disks cost approximately CNY 11,200 per year.

  • You must handle ECS supply-chain constraints during peak traffic and idle resources during off-peak hours, which may lead to resource waste.

Network fees

  • Internal network access within the same region is free of charge.

  • Cross-region Internet access is free of traffic charges. Alibaba Cloud Elasticsearch provides Internet access for free, with a default bandwidth of 5 Gbps.

  • Internal network access within the same region is free of charge.

  • Cross-region Internet access requires you to pay for the outbound traffic yourself. For detailed pricing, see Public bandwidth.

Labor and time

  • Alibaba Cloud ELK (Elasticsearch, Logstash, and Kibana) is fully managed, ready to use out of the box, and billed on demand.

  • Visual, GUI-based cluster O&M and management capabilities keep operational overhead extremely low.

  • You must purchase machines and build the Elasticsearch cluster manually. Iteration is slow and time costs are high.

  • You must manage resources and perform cluster O&M yourself, which requires a dedicated Elasticsearch engineering team and results in high labor costs.

Risk cost

The cloud service guarantees 99.9% service reliability, keeps IT risk to a minimum, and makes upper-layer business risk manageable.

A self-managed Elasticsearch cluster provides no minimum reliability guarantee. Reducing business risk requires strong technical expertise and significant investment.

Feature cost

  • All advanced features of the commercial X-Pack plug-in are included for free.

  • Free OSS-based data backup is provided.

  • You must pay for the commercial X-Pack suite, which is valued at more than CNY 36,000.

  • You must back up data yourself, and backup storage is billed separately.

Item

Alibaba Cloud Elasticsearch

Self-managed Elasticsearch on ECS

Usability

  • Deployment is complex and resources must be adjusted manually.

  • Data must be migrated before a cluster version upgrade.

  • O&M is difficult and requires running commands to check the health of clusters, nodes, indexes, and more.

  • Provides only basic monitoring capabilities. Complex metrics require secondary encapsulation.

Scenario support

  • All advanced commercial X-Pack features are included for free.

  • Supports scenario-based configuration templates that provide optimized parameter settings.

  • Search scenario capabilities: an NLP word segmentation plug-in from Alibaba DAMO Academy, a vector search plug-in, and a self-developed SQL plug-in are provided.

  • Log scenario capabilities: the high-performance Indexing Service delivers read/write separation, and Openstore intelligent hybrid storage delivers storage-compute separation, enabling low-cost storage of massive log volumes. For more information, see AliES kernel-enhanced edition.

Capabilities must be developed in-house or integrated from open source.

Performance

  • Based on the enhanced kernel, a read/write-separation and storage-compute-separation architecture delivers higher performance at lower cost.

  • More than 10 deeply customized kernel-enhancement engine features improve cluster read and write performance.

You must ensure performance yourself, which is technically difficult to implement.

Availability

  • Supports automatic data backup.

  • Data and service reliability reach 99.9%.

  • Backed by the cloud-native, self-developed AliES enhanced kernel engine, features such as a cluster throttling plug-in and slow query isolation ensure cluster stability.

  • Supports multi-zone deployment, providing an active zone-redundancy architecture across zones in the same city.

  • You must back up data manually and ensure cluster availability yourself.

  • Implementing disaster-recovery deployment is technically difficult.

Security

  • By default, an Elasticsearch cluster is accessed within a virtual private cloud (VPC).

  • X-Pack security components are included for free.

  • Field-level access control is supported.

  • Supports HTTPS-based encryption in transit and encryption at rest.

  • Only basic ECS-level security is provided, so a self-managed Elasticsearch cluster carries relatively high security risk.

  • X-Pack security components must be purchased separately.