Skip to main content
Kuldeep Singh
Kuldeep SinghDigital Architecture
SALESFORCE · LWC · AWS · SEARCH INFRASTRUCTURERole: Principal Systems Architect

High-Volume Multi-Site Quoting

Compressing the payload. Moving computation closer to the experience.

Compressing the payload. Moving computation closer to the experience.

LWCPAKO/GZIPAWSRUSTTYPESCRIPT
01 // The Constraint

What Made the Problem Difficult

High-volume multi-site quoting in telecommunications and connectivity infrastructure involves configuring hundreds or thousands of physical site locations, complex bandwidth tiers, and dynamic product catalogs within a single commercial negotiation.

Core Platform Constraint

The quoting experience had to remain a native Salesforce Lightning Web Component (LWC) interface. However, transferring and processing massive quoting datasets overwhelmed browser memory, caused UI thread freezing, and collided with platform payload and heap limits.

Boundary Invariants & Operational Limits:
  • !Quoting experience must remain strictly native within Salesforce Lightning Web Components (LWC)
  • !Prevent browser tab crashes and UI thread freezes during massive multi-site payload ingestion
  • !Operate reliably within Salesforce platform transport limits and LWC security boundaries
  • !Do not force sales reps or partners out of Salesforce into external standalone web portals
  • !Enable lightning-fast search and location filtering across large commercial catalog datasets
02 // The Architecture

What System Was Designed

Architected an end-to-end data compression and distributed search retrieval pipeline. The native LWC client leverages the pako library via platformResourceLoader to compress payloads in browser memory. Compressed streams travel across an AWS supporting transport layer that orchestrates high-volume processing and interfaces with dual search engines: a high-throughput Rust search engine for deep indexing, and an agile TypeScript search engine for dynamic in-memory filtering. Targeted results hydrate incrementally back into the native Salesforce UX.

03 // The Key Decision

Pivotal Architectural Choices

Key Architectural Decisions

ADR // 01DETERMINISTIC CHOICE

Client-Side Binary Compression (pako) vs. Off-Platform Redirection

Context:

Industry convention redirects high-volume users to off-platform external portals when payloads exceed CRM tolerances.

Alternatives Evaluated:
  • Offload quoting entirely to standalone external web app
  • Heavy server-side Apex pagination with dozens of queries
  • Uncompressed REST batch polling
Chosen: In-browser gzip compression via pako loaded through platformResourceLoader coupled with an AWS transport layer.

Preserves native Salesforce workflows and CRM context for account executives while shrinking data payloads over the wire by up to 80-90%.

ADR // 02DETERMINISTIC CHOICE

Dual-Engine Retrieval Topology (Rust + TypeScript)

Context:

Location filtering requires heavy index scanning across vast geographic data, while interactive quoting requires dynamic UI-level adjustments.

Alternatives Evaluated:
  • Single monolithic database query
  • Pure client-side array filtering in LWC
Chosen: Rust search engine for high-volume indexing + TypeScript search engine for dynamic evaluation.

Assigns each workload to its ideal runtime: Rust provides zero-overhead index retrieval, while TypeScript delivers rapid in-memory rule adjustments.

04 // The Engineering

What Was Actually Built

01

Integrated the pako gzip compression library inside Salesforce LWC static resources using lightning/platformResourceLoader.

02

Engineered binary array-buffer streaming across network boundaries to bypass JSON serialization bloat.

03

Built an AWS supporting transport layer handling stream ingestion, multi-site chunking, and worker orchestration.

04

Developed a dedicated Rust search engine service delivering high-throughput spatial and attribute indexing.

05

Implemented a complementary TypeScript search engine for agile client/edge filtering and ranking.

06

Constructed an incremental hydration virtual DOM renderer in LWC to ensure buttery 60fps scrolling across thousands of quote lines.

05 // The Trade-offs

Deliberate Architectural Compromises

Trade-offs & Mitigations

Browser Decompression CPU Overhead vs Network Transfer Time

Architectural Benefit:

Immunity to network payload caps and dramatic reduction in transit latency.

Associated Cost:

Client browser CPU expends cycles decompressing binary payloads.

Mitigation Strategy:

Chunked decompression scheduled via micro-task yielding to prevent blocking user interaction frames.

Specialized Dual-Language Stack (Rust/TS) vs Monolithic Framework

Architectural Benefit:

Extreme indexing throughput from Rust combined with rapid UI agility from TypeScript.

Associated Cost:

Engineering governance across multiple language ecosystems.

Mitigation Strategy:

Strict contract-first protocol specifications governing data serialization between services.

06 // The Result

Verified Outcomes

✓ VERIFIED RESULT

Preserved 100% native Salesforce LWC quoting experience without external portal redirects

✓ VERIFIED RESULT

Eliminated browser memory exhaustion and tab crashes during high-volume multi-site quoting

✓ VERIFIED RESULT

Engineered robust end-to-end data transport handling massive multi-site quoting datasets

✓ VERIFIED RESULTRequires Benchmark Confirmation

Specific throughput, site volume, and compression metrics: [VERIFY WITH KULDEEP]

07 // Architecture Blueprint

System Schematic & Data Flow

System Topology Blueprint
Connectbase High-Volume Quoting & Search Topology
INTERACTIVE SCHEMATIC
SALESFORCE PLATFORM BOUNDARY · NATIVE EXPERIENCENative LWC Quoting UIpako Gzip Compression CoreIncremental Row Hydrationcompressed payload stream (gzip)AWS SUPPORTING TRANSPORT & ORCHESTRATION LAYERStream Ingestion, Multi-Site Chunking & Workload DistributionSEARCH ENGINE A · HEAVY INDEXINGRust Search EngineHigh-throughput location index & spatial searchSEARCH ENGINE B · DYNAMIC EVALUATIONTypeScript Search EngineAgile in-memory filtering & rule evaluationtargeted minimal result set
PRESERVED NATIVE UXSales and operations stay inside native Salesforce LWC without external portal redirects.
CLIENT-SIDE GZIPPayloads compressed via pako in browser memory before network transit.
TARGETED HYDRATIONOnly resolved, minimal records return to the DOM, shielding UI thread from memory freezes.

Native LWC uses pako gzip compression, AWS transport, and dual Rust/TS search engines to handle massive multi-site quoting datasets.

Text alternative for screen readers: Architecture flow: Native LWC UI to pako Gzip Engine via Client In-Memory Compression; pako Gzip Engine to AWS Transport Layer via Compressed Binary Stream; AWS Transport Layer to Rust & TS Engines via Dual Search Execution; Rust & TS Engines to Native Salesforce UX via Targeted Result Hydration

08 // Why It Matters

Architectural Conclusion

A Salesforce-native experience can remain native without forcing Salesforce or the browser to carry every byte of a high-volume quoting problem.

Key Lessons for Platform Scale:
  • Architecting around platform constraints does not mean abandoning the platform; it means being intentional about where data is compressed, where it travels, and where it is computed.
  • Native LWC can be supercharged with battle-tested libraries through static resources to conquer massive data volumes.