Why a Domestic Elasticsearch Fork Won Major Bank & Insurance Contracts: Five Hard-Won Factors

Analysis of INFINI Labs' Easysearch securing Bank of China and PICC contracts, highlighting full Elasticsearch API compatibility, kernel-level enterprise security, certified indigenous hardware adaptation, direct vendor operations support, and the strategic shift of search engines into AI knowledge bases.

Mingyi World Elasticsearch
Mingyi World Elasticsearch
Mingyi World Elasticsearch
Why a Domestic Elasticsearch Fork Won Major Bank & Insurance Contracts: Five Hard-Won Factors

01 Compatibility as a Prerequisite, Not a Bonus

Enterprises have built business code, SDKs, DSL queries, and Kibana dashboards on Elasticsearch for years; rewriting them is a blocker. Easysearch addresses this by providing full compatibility with REST APIs, Java/Python/Go/Node.js SDKs, Query DSL, and ES 6.x/7.x/8.x data read/write. Kibana, Logstash, Beats, and Filebeat connect without changes. This enables zero-rewrite migration — a make-or-break factor for the Bank of China project, which handles hundreds of millions of daily writes, hundreds of terabytes, and hundreds of cross-region nodes. The support scope also includes migration from x86 to domestic CPUs (Kunpeng, Phytium) and OSes (UOS, Kylin), clearing the infrastructure path first.

02 Enterprise-Grade Capabilities Built Into the Kernel

Many domestic alternatives pass proof-of-concept but fail in production because security and governance features are bolted on externally. Easysearch embeds TLS encryption, RBAC/ABAC permissions, field-level masking, LDAP/AD authentication, IP allow/block lists, anti-brute-force, snapshot encryption, rate limiting, audit logging, and national cryptographic algorithms directly in the kernel. These are hard requirements for passing financial and government security reviews. The PICC win illustrates this: their pain point was fragmented ES cluster versions, uncontrolled traffic, and mounting operational burden. INFINI Gateway provides a unified control plane for multi-version clusters (ES 2.x–8.x, OpenSearch 2.x), index- and user-level rate limiting, cross-cluster governance, and full audit with dynamic masking — delivering a manageable infrastructure, not just a query engine.

03 Indigenous Innovation Adaptation Validated by Hardware Vendor Joint Testing

Superficial "Xinchuang" adaptation often means a one-off demo on a domestic chip that collapses under real load. True adaptation requires joint testing with hardware vendors. Easysearch, Gateway, Console, and Coco AI completed dual-architecture (x86 and ARM) testing on FusionOS 26 with dual-party sign-off. Combined with prior certifications on Loongson, Kunpeng, Phytium, Kylin, and UOS, coverage spans mainstream domestic hardware/software stacks. At DTCC 2026, the speaker outlined five selection dimensions — self-control, enterprise stability, real Xinchuang adaptation, security governance, and long-term evolution — noting that Xinchuang adaptation tests willingness to do the "hard work." With Elastic's official withdrawal from this space and other domestic vendors lagging, this gap represents both risk and opportunity.

04 Direct Vendor Support Determines Long-Term Viability

Selection is only the start; success hinges on 3–5 years of operations and troubleshooting. The Bank of China support contract includes cluster health checks, root-cause analysis (node isolation, circuit breaking), JVM and index tuning, slow-query elimination, and 7×24 high availability. Troubleshooting and tuning rely on deep kernel understanding, not manual recitation. Original-vendor direct support provides a depth that third parties cannot match.

05 The Strategic Shift: Search Engines as AI Knowledge Bases

Search engines have moved from auxiliary query tools to core infrastructure for RAG and Agent workloads, carrying enterprise core data. Failure now means total system compromise, turning "self-control" from optional to mandatory. Easysearch's path: lower migration cost to a board-approvable level, embed enterprise capabilities in the kernel rather than as add-ons, then build trust through real joint testing and hands-on service. Missing any link breaks the chain — compatibility without enterprise features crashes in production; enterprise features without Xinchuang adaptation cannot enter the door; adaptation without sustained service cannot maintain the relationship. Domestic substitution is not merely swapping an engine; it is building a foundation for the next several years. These wins demonstrate that the hard work has been done.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

XinchuangElasticsearch compatibilityEasysearchARM x86 dual architectureBank of ChinaFusionOSindigenous innovationPICCRAG knowledge base
Mingyi World Elasticsearch
Written by

Mingyi World Elasticsearch

The leading WeChat public account for Elasticsearch fundamentals, advanced topics, and hands‑on practice. Join us to dive deep into the ELK Stack (Elasticsearch, Logstash, Kibana, Beats).

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.