Consumers Clients
Web and mobile apps, partner systems, internal callers.
request over HTTPS
Enterprise Integration
Enterprise Integration
Quince Software delivers enterprise integration, cloud-native on AWS, IBM Cloud, Azure or GCP, on-premises or hybrid, with twelve-plus years behind it. Architecture, build, security integration design, security compliance, and the running of estates other people depend on.
Enterprise authorization is the firm's other discipline and has its own page. Nothing from it has been applied inside an integration estate.
The shape of the work
One front door, many systems of record
Wider than the screen. Scroll the diagram sideways to follow it.
Reference architecture · integration
One runtime sits behind the gateway and managed file transfer
Backend services
Kubernetes or OpenShift on AWS, IBM Cloud, Azure or GCP · Backend microservices · MQ and Kafka as a managed service or in-cluster
API gateway Edge
One front door: routing, authentication and quotas.
routed to a service
Managed file transfer Files
Partner file drops and scheduled transfers.
batch received
Microservices ACE
One container per capability, all behind the gateway.
/orders · /invoices · /payments · /customers · /assortments · /pricing · /shipments
MQ Queues
Transactional messages, delivered once and in order.
message queued
Apache Kafka Events
Event streams, replayable, for downstream consumers.
event published
Datastore State
Correlation keys, in-flight status and cached reference data.
CouchDB, or the client's own
SAP S/4HANA ERP
Finance, sales and distribution.
IDoc, JCo, BAPI
Salesforce CRM
Accounts, opportunities and cases.
REST API
E-commerce Shop
Orders, catalogue and stock from the storefront.
webhooks and REST
Payment and settlement Finance
Settlement files into the finance ledger.
files and messages
What we deliver
Architecture through to operations
Integration architecture and delivery
Architecture through to running code on the IBM stack: App Connect Enterprise and Integration Bus, MQ, API Connect and Cloud Pak for Integration, with Apache Kafka where the traffic is event-shaped.
Cloud, serverless and containers
Integration built on cloud platforms rather than on an appliance: AWS, IBM Cloud, Azure or GCP, with serverless functions, API gateways and event-driven flows, and containerised App Connect Enterprise runtimes on Kubernetes or OpenShift.
SAP integration
IDoc, JCo and BAPI interfaces, SD and retail flows, finance settlement interfaces, and e-commerce and payment integrations into SAP finance.
Security integration design
How an estate holds its credentials, secures traffic in transit and at rest, authenticates callers at the gateway, separates its environments, and what evidence the design leaves behind for an auditor to read.
Running it, or leading the team that does
Managed integration services under an agreed SLA, managed file transfer modernisation, and integration team leadership, currently a six-engineer team at a large retailer.
Industries served
- Consumer goods
- Retail
- Banking
- Automotive
- Pharma
- Telecom
- Petroleum
Scale
What the current platform carries
| Measure | Result |
|---|---|
| Applications in the estate | ~100, on Kubernetes and OpenShift, serverless functions and an API gateway |
| API gateway traffic | ~150,000 calls a week |
| Orders through the platform | ~100,000 a week |
| Availability | 99.9% |
Measured on the integration platform of the engagement the firm currently leads.
How a solution is built
Layers, deployment models and patterns
Layers
Edge to adapters follow the diagram above, top to bottom; platform, security and operations sit beneath them.
| Layer | What it does |
|---|---|
| Edge | Admits every call to the estate through one gateway, and every partner file through managed file transfer. |
| Services | Maps the data passing between consumers and systems of record into the form each side expects, with one service for each business capability. |
| Messaging | Carries transactional messages on queues that neither lose nor reorder them, and events on streams that any number of downstream systems can read. |
| State | Keeps track of each flow from its first call to its last reply, and caches the reference data it looks up. |
| Adapters | Reaches each system of record through the interface it already offers: its own protocol, a REST API, webhooks, files or messages. |
| Platform | Runs the services in containers on an orchestrated cluster or as serverless functions, with the messaging and the datastore in the same cluster or as managed services. |
| Security | Applies at every layer, from the caller authenticated at the gateway to the credentials each adapter holds for its system of record. |
| Operations | Logs the transactions passing through the estate, the record the team running it works from. |
Deployment models
Cloud-native
Every integration layer runs on a public cloud, the services in containers or as serverless functions.
AWS · IBM Cloud · Azure · GCP
On-premises
The same layers, in the client's own data centre.
Hybrid
Some layers run on a public cloud and the rest in the client's own data centre.
Architectural patterns
Each card names the parts of the diagram above that carry the pattern.
Microservices
Orders, payments, pricing and the other business capabilities each run as a separate service with its own endpoint.
- Microservices
Request and reply
The caller's HTTPS call passes the gateway to one service, and the caller waits for that service's answer.
- Consumers
- API gateway
- Microservices
Guaranteed messaging
A transactional message waits on its queue until the receiving system has taken it.
- MQ
Publish and subscribe
Every system subscribed to a stream reads each event published to it, and can replay the stream later.
- Apache Kafka
File-based batch
Partner and settlement files travel whole, on a schedule, and are processed as a batch.
- Managed file transfer
- Payment and settlement
Correlation identifier
When a reply arrives later, a key kept in the datastore matches it to the request that started it.
- Datastore
Compliance
Start with the estate you have
An integration engagement starts with what is already running, what it costs to keep running, and what has to change first.