Tables.java

/**
 * Copyright 2015-2016 The OpenZipkin Authors
 *
 * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
 * in compliance with the License. You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software distributed under the License
 * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
 * or implied. See the License for the specific language governing permissions and limitations under
 * the License.
 */
package zipkin.storage.cassandra;

import zipkin.Endpoint;
import zipkin.storage.QueryRequest;
import zipkin.storage.SpanStore;

final class Tables {

  /**
   * This index supports {@link SpanStore#getServiceNames()}}.
   *
   * <p>The cardinality of {@link Endpoint#serviceName} values is expected to be stable and low.
   * There may be hot partitions, but each partition only includes a single column. Hence bucketing
   * would be unnecessary.
   */
  static final String SERVICE_NAMES = "service_names";

  /**
   * This index supports {@link SpanStore#getSpanNames(String)}.
   *
   * <p>The compound partition key includes {@link Endpoint#serviceName} and a constant bucket (0).
   * This is because span names are bounded (and will throw an exception if aren't!).
   */
  static final String SPAN_NAMES = "span_names";

  /**
   * This index supports trace id lookups by {@link QueryRequest#serviceName}, within the interval
   * of {@link QueryRequest#endTs} - {@link QueryRequest#lookback}.
   *
   * <p>The cardinality of {@link Endpoint#serviceName} values is expected to be stable and low. To
   * avoid hot partitions, the partition key is {@link Endpoint#serviceName} with abucket (random
   * number between 0 and 9).
   */
  static final String SERVICE_NAME_INDEX = "service_name_index";

  /**
   * This index supports trace id lookups by {@link QueryRequest#serviceName} and {@link
   * QueryRequest#spanName}, within the interval of {@link QueryRequest#endTs} - {@link
   * QueryRequest#lookback}.
   *
   * <p>The partition key is "{@link Endpoint#serviceName $serviceName}.{@link zipkin.Span#name
   * $spanName}", which is expected to be diverse enough to not cause hot partitions.
   */
  static final String SERVICE_SPAN_NAME_INDEX = "service_span_name_index";

  /**
   * This index supports trace id lookups by {@link QueryRequest#annotations} or {@link
   * QueryRequest#binaryAnnotations}, within the interval of {@link QueryRequest#endTs} - {@link
   * QueryRequest#lookback}.
   *
   * <p>The annotation field is a colon-delimited string beginning with {@link
   * Endpoint#serviceName}. If an {@link zipkin.Annotation}, the second part will be the value. If a
   * {@link zipkin.BinaryAnnotation}, the second part would be the key, and the third, the value.
   *
   * <p>For example, an annotation of "error" logged by "backend2" would be stored as the annotation
   * field "backend2:error". A binary annotation of "http.method" -> "GET", logged by "edge1" would
   * result in two rows: one with annotation field "edge1:http.method" and another with
   * "edge1:http.method:GET".
   *
   * <p>To keep the size of this index reasonable, {@link zipkin.Constants#CORE_ANNOTATIONS} are not
   * indexed. For example, "service:sr" won't be stored, as it isn't supported to search by core
   * annotations. Also, binary annotation values longer than 256 characters are not indexed.
   *
   * <p>Lookups are by equals (not partial match), so it is expected that {@link zipkin.Annotation}
   * and {@link zipkin.BinaryAnnotation} keys and values will be low or bounded cardinality. To
   * avoid hot partitions, the partition key is the annotation field with a bucket (random number
   * between 0 and 9).
   */
  static final String ANNOTATIONS_INDEX = "annotations_index";

  private Tables() {
  }
}