Whats New

What's New in Elastic MDS

This page is for existing MetaFluent v5 customers. It sets out what Elastic MDS adds beyond the v5 platform you already run.

Audience: existing MetaFluent v5 customers.

v5 application code is fully compatible and runs unchanged. The v5 compatibility details are in JMS Application Development.

---

Automatic slow-consumer management

A distribution server fans the same updates out to many independent consumer connections. When one consumer - its application, its network, or a downstream system - cannot keep pace, its unread data backs up on the server and, left unmanaged, degrades delivery for the other consumers sharing that server. Elastic MDS manages this automatically, within the standard distribution infrastructure:

• No code changes. Slow-consumer handling is a property of the platform. Applications adopt it by doing nothing.
• Transparent to applications. A consumer that falls behind is kept current without being disconnected and without an error it has to handle.
• One infrastructure. There is no separate slow-consumer tier or add-on product to deploy and operate. The same distribution servers that deliver your content also protect it.

The protection is multi-level. The platform continuously profiles each consumer connection and responds in order of least disruption first:

• Split before conflate. Where it can, the platform first splits a consumer's stream across additional connections, spreading the load so the consumer is served within capacity. This mitigation is transparent and carries no data loss.
• Conflate only when it must. A connection that still cannot sustain the offered rate has conflation engaged for that connection alone: updates arriving in a short window are coalesced into a single update carrying the current value. The slow consumer receives fewer messages but always the latest data - never stale, never silently dropped - and consumers keeping up are left untouched. Conflation is disengaged automatically when the connection recovers.

The mechanism, including how each connection's capacity ceiling is learned rather than fixed, is described in Architecture: Advanced.

---

JMS selectors

Elastic MDS implements the JMS message selector for content subscriptions. A subscription can carry a selector that controls what the application receives, so an application sees only the fields it wants - which reduces the bandwidth consumed, since both image and update messages carry fewer fields, and the CPU used on the client.

• A very minor application change. Adopting a selector is a single string passed when the subscriber is created; the rest of the application is unchanged.
• Selectors work under the v5 context. A v5 application on com.metafluent.jms_context.mds can adopt them without moving to the new MarketData context.

A selector does more than choose fields - it can rename a field with AS, compute a derived field from an expression, and constrain the update stream with a WHERE clause so an update is delivered only when a condition holds. The full grammar and examples are in JMS Application Development.

---

System-wide monitoring and control

Elastic MDS gives an operator a centralized, fine-grained line of sight to the entire deployment, regardless of how many containers it spreads across. Each component reports its health, role, and live statistics through one API, which drives both a dashboard and direct programmatic access:

• REST API. Each operational query - component health, active/standby roles, connection and consumer statistics, whether conflation is currently engaged on a connection - is available over a standard REST interface, addressed through the API gateway. It is the control and observation interface for the whole cluster, and the dashboard is built on it. See Monitoring & Diagnostics and the REST API reference.
• Dashboard. A graphical Dashboard presents the same system-wide state visually, for operators who prefer that to issuing queries directly.

---

Where to go next

• Architecture: Basics - the model these capabilities build on, then Architecture: Advanced for the deployment, resilience, and slow-consumer detail.
• JMS Application Development - selectors, contexts, and how v5 code carries over.
• Monitoring & Diagnostics - seeing a running deployment through the gateway.
• Quick Start - stand up a deployment and watch data flow.