The platform behind your next product.

01 / THE DATA MODEL

A Dataset defines what you cover.

Which entities do you have information about, and what can your product do with it? A Dataset brings together a Catalog and, optionally, one canonical Stream for the entities it covers. The Catalog describes those entries; the Stream carries their time-ordered updates.

Each Dataset has its own coverage: a broad instrument universe, a venue’s listings, or a specialist set of structured-product terms. Coverage is actual membership—not a category label, and not a promise that every field or historical period is populated.

02 / CONNECT THE IDENTIFIERS

Different identifiers. The same entity.

One provider uses an ISIN, another a WKN, a third its own identifier. Connect their information through shared identity resolution instead of rebuilding cross-provider mapping tables in every application.

Identity connects. Coverage tells you what is available. Knowing that two identifiers refer to the same instrument does not mean every Dataset contains it. Combine identity resolution with remixed Dataset coverage to join provider information, discover specialist data, and offer the right product features for that entity.

03 / REMIX YOUR UNIVERSE

Your main Dataset. A map to everything else.

Remixing is how you build the Dataset your product starts from: the entities you cover, the information you bring together, and the specialist data your users can explore next.

04 / THE SERVICES

One Gateway. Purpose-built services.

05 / CONNECT

Your feeds. On a shared foundation.

You should choose feeds for your product—not to fit our connector list. We implement or adapt the parsers, protocol handling, and mappings for the providers you need, using the shared ingestion, recovery, storage, and delivery infrastructure already in place. Your pilot defines the exact interfaces and the integration work still required.

06 / ENGINEERING TARGET

Two million instruments.
One fast CPU core.

07 / BUILD

Typed data. Familiar code.

Schema-generated SDKs carry your fields into your developers’ workflow. This example from our generated Globex SDK reads a quote sample with typed bid/ask, trade, and custom fields. Exchange messages stay grouped; prices remain exact decimal strings.

import { connect, selector, type ConnectOptions } from "@globex/market-data";

export async function readQuoteSample(options: ConnectOptions, symbol: string) {
  const connection = await connect(options);
  try {
    const requested = selector.us(symbol).at("IEX");
    const result = [];
    for await (const msg of connection.stream({
      selector: requested,
      blocks: ["BidAsk", "Trade", "ClientTradeAttributes"],
    })) {
      if (msg.selector !== requested) throw new Error("response belongs to another selector");
      console.log("yes, it is what we requested");
      result.push({
        phase: msg.phase,
        requestedSelector: msg.selector.expression,
        sourceRecord: msg.sourceRecord,
        messageId: msg.messageId.toString(),
        bid: msg.fields.bidAsk?.bid?.price.toString() ?? null,
        ask: msg.fields.bidAsk?.ask?.price.toString() ?? null,
        trade: msg.fields.trade?.price.toString() ?? null,
        clientTradeAttributes: msg.fields.clientTradeAttributes ?? null,
        presentFields: [...msg.fieldIterator()].map((field) => field.name),
      });
      if (result.length === 2) break;
    }
    return result;
  } finally {
    await connection.close();
  }
}

08 / OPERATE

Durable storage. Replaceable services.

OpenMDTA’s durable state lives in S3-compatible storage, not in an irreplaceable service instance. Restart or replace services, reconnect them to the same storage, and resume from persisted state. Reliable storage is the foundation of recovery—without an OpenMDTA-specific backup-and-restore procedure for each service.

Keep deployment configuration and credentials available independently. Storage must provide the required consistency and conditional-write semantics; availability, retention, and data protection remain part of the deployment design.