...
This page is meant as a template for writing a KIP. To create a KIP choose Tools->Copy on this page and modify with your content and replace the heading with the next KIP number and a description of your issue. Replace anything in italics with your own description.
Status
Current state: Under Discussion Accpted
Discussion thread: here
JIRA: here :https://lists.apache.org/thread/ogo7ntmj8srdcko2h86vvd9djjsjfvcj
Vote thread: https://lists.apache.org/thread/q4kn2g6tmc837ph2zvff40pgpmgzok3d
JIRA:
Jira | ||||||
---|---|---|---|---|---|---|
|
Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).
Motivation
In the current IQv2 code, there are noticeable differences when interfacing with plain- kv-store
and ts-kv-store
. Notably, the return type V
acts as a simple value for plain- kv-store
but evolves into ValueAndTimestamp<V>
for ts-kv-store
, which presents type safety issues in the API.
Even if IQv2 hasn't gained widespread adoption, an immediate fix might bring compatibility concerns.
This brings us to the essence of our proposal: the introduction of distinct query types. One that returns a plain value, another for values accompanied by timestamps.
Although it's viable to issue a "plain value" query against a ts-kv-store and then unpack the return value, doing the reverse — launching a ValueAndTimestamp<V> query against a plain-kv-store — seems illogical.
Our vision is that plain-kv-store should always return a plain V. However, for ts-kv-store, there's a pronounced need for specialized query types. Ultimately, the user should only be required to define V.
To address these concerns, we propose:
Code Block | ||
---|---|---|
| ||
public final class KeyQuery<K, V> implements Query<V>
public final class TimestampKeyQuery<K, V> implements Query<ValueAndTimestamp<V>> |
Why introduce TimestampKeyQuery and TimestampRangeQuery? The primary motivation behind this is to ensure type safety and foster a clear distinction in our API. They bridge the difference between simple key-value stores and those integrated with timestamps, offering a more robust and intuitive querying mechanism.
Proposed Changes
Within the current IQv2 codebase, there have been distinct interactions between plain-kv-store and ts-kv-store. These differences, especially in return types, have raised concerns over type safety within the API.
To address these challenges and streamline the querying experience, we have decided to refine our approach and introduce two specialized query types: TimestampKeyQuery and TimestampRangeQuery.
TimestampKeyQuery: This query type will consistently return ValueAndTimestamp<V>, ensuring that there's a clear and predictable return type associated with timestamped key-value stores.
Code Block | ||||
---|---|---|---|---|
| ||||
@Evolving
public final class TimestampKeyQuery<K, V> implements Query<ValueAndTimestamp<V>> {} |
TimestampRangeQuery: Tailored for ranges with timestamps, this query will return a KeyValueIterator<K, ValueAndTimestamp<V>>
Code Block | ||||
---|---|---|---|---|
| ||||
@Evolving
public final class TimestampRangeQuery<K, V> implements Query<KeyValueIterator<K, ValueAndTimestamp<V>>> {} |
Previously, MeteredKeyValueStore was equipped to handle both plain V queries and ValueAndTimestamp<V> queries. With this update, all KeyQuery instances will only return the plain V, eliminating the previously supported ValueAndTimestamp<V>. On the other hand, all TimestampKeyQuery instances are now designed to strictly return ValueAndTimestamp<V>.
This restructuring ensures a more intuitive, type-safe, and consistent querying mechanism for users across different types of key-value stores in the IQv2.
Compatibility, Deprecation, and Migration Plan
- Utilizing the existing RangeQuery and KeyQuery class, we can make some modifications to realize the concepts of TimestampKeyQuery and TimestampRangeQuery.
- Since nothing is deprecated in this KIP, users have no need to migrate unless they want to.
Test Plan
To ensure the robustness and accuracy of our new query types, TimestampKeyQuery and TimestampRangeQuery, it's essential to have thorough test coverage. With that in mind, we propose the creation of two specific test methods:
shouldHandleTimestampKeyQuery: This test method will validate the functionality of TimestampKeyQuery, ensuring it consistently returns ValueAndTimestamp<V> as expected.
| ||||
Code Block | ||||
---|---|---|---|---|
| ||||
public <V> void shouldHandleTimestampKeyQueryshouldHandleKeyQuery( final Integer key, final Function<ValueAndTimestamp<V>Function<V, Integer> valueExtactor, final Integer expectedValue) { ... final TimestampKeyQuery<IntegerKeyQuery<Integer, V> query = TimestampKeyQueryKeyQuery.withKey(key); ... final StateQueryRequest<ValueAndTimestamp<V>>StateQueryRequest<V> request = inStore(STORE_NAME) .withQuery(query) .withPartitions(mkSet(0, 1)) .withPositionBound(PositionBound.at(INPUT_POSITION)); ... final StateQueryResult<ValueAndTimestamp<V>>StateQueryResult<V> result = IntegrationTestUtils.iqv2WaitForResult(kafkaStreams, request); ... final QueryResult<ValueAndTimestamp<V>>QueryResult<V> queryResult = result.getOnlyPartitionResult(); ... final ValueAndTimestamp<V>V result1 = queryResult.getResult(); final Integer integer = valueExtactor.apply(result1); assertThat(integer, is(expectedValue)); ... } |
Before the introduction of TimestampedKeyQuery
, when using KeyQuery
, we obtained the result using the following code:
Code Block |
---|
final V result1 = queryResult.getResult(); |
This meant that the returned result could be of two potential data types: plain V
or ValueAndTimestamp<V>
. This was a source of inconsistency. For instance, querying a kv-store
with KeyQuery
would return a V
type, but querying a ts-kv-store
would yield a ValueAndTimestamp<V>
. This behavior is unintuitive and potentially confusing for developers.
To ensure consistency, we suggest that KeyQuery always return the plain V type, enhancing the predictability of the mentioned code. Likewise, RangeQuery should uniformly return the plain V KeyValueIterator.
For those requiring timestamped values from a ts-kv-store
, we recommend introducing a new query type: TimestampedKeyQuery
. This new query will specifically target ts-kv-stores
and will return ValueAndTimestamp<V>
. Furthermore, to complement this, TimestampedRangeQuery
should be introduced to query ranges in ts-kv-stores
, ensuring that the returned value always includes timestamps.
Code Block | ||||
---|---|---|---|---|
| ||||
public final class TimestampedKeyQuery<K, V> implements Query<ValueAndTimestamp<V>> |
Code Block | ||||
---|---|---|---|---|
| ||||
public final class TimestampedRangeQuery<K, V> implements Query<KeyValueIterator<K, ValueAndTimestamp<V>>> |
Why introduce TimestampedKeyQuery
and TimestampedRangeQuery
? The primary motivation behind this is to ensure type safety and foster a clear distinction in our API. They bridge the difference between simple key-value stores and those integrated with timestamps, offering a more robust and intuitive querying mechanism.
Proposed Changes
Within the current IQv2 codebase, there have been distinct interactions between plain-kv-store and ts-kv-store. These differences, especially in return types, have raised concerns over type safety within the API.
To address these challenges and streamline the querying experience, we have decided to refine our approach and introduce two specialized query types: TimestampedKeyQuery
and TimestampedRangeQuery
.
TimestampedKeyQuery
: This query type will consistently return ValueAndTimestamp<V>
, ensuring that there's a clear and predictable return type associated with timestamped key-value storesshouldHandleTimestampRangeQuery: This method is tailored to verify the TimestampRangeQuery, ensuring that it correctly returns a KeyValueIterator<K, ValueAndTimestamp<V>>.
Code Block | ||||
---|---|---|---|---|
| ||||
private@Evolving public <T>final void shouldHandleTimestampRangeQueries(final Function<ValueAndTimestamp<T>, Integer> extractor) { shouldHandleTimestampRangeQuery( class TimestampedKeyQuery<K, V> implements Query<ValueAndTimestamp<V>> { ... /** * Creates a query that will retrieve the record identified by {@code key} if it exists * Optional.of(0),(or {@code null} otherwise). * @param key The key to retrieve Optional.of(4), * @param <K> The type of the key * @param <V> The type of the value extractor,that will be retrieved */ public static mkSet(1<K, V> TimestampedKeyQuery<K, 3, 5, 7, 9) V> withKey(final K key) /** * Specifies ); that the cache should be skipped during query ... } |
We will focus on conducting a detailed test for shouldHandleTimestampRangeQuery.
Code Block | ||||
---|---|---|---|---|
| ||||
public <V> void shouldHandleTimestampRangeQuery( final Optional<Integer> lower,evaluation. This means, that the query will always * get forwarded to the underlying store. */ public TimestampedKeyQuery<K, V> skipCache() final/** Optional<Integer> upper, * The key that was specified for this query. final Function<ValueAndTimestamp<V>, Integer> valueExtactor,*/ public K key() /** final Set<Integer> expectedValue) { * The flag whether to skip the cache finalor TimestampRangeQuery<Integer,not V>during query; evaluation. */ public boolean isSkipCache() |
TimestampedRangeQuery
: Tailored for ranges with timestamps, this query will return a KeyValueIterator<K, ValueAndTimestamp<V>>
According to KIP-968, this KIP introduces the public enum ResultOrder to determine whether keys are sorted in ascending or descending or unordered order. Order is based on the serialized byte[] of the keys, not the 'logical' key order. employs the withDescendingKeys() and withAscendingKeys()
methods to specify that the keys should be ordered in descending or ascending or unordered sequence, and the resultOrder() method to retrieve the value of enum value in ResultOrder. I've incorporated these variables and methods into the TimestampedRangeQuery
class and modified some method inputs. As a result, we can now use withDescendingKeys()
to obtain results in reverse order and use withAscendingKeys to obtain the result in ascending order.
Code Block | ||||
---|---|---|---|---|
| ||||
@Evolving public final class TimestampedRangeQuery<K, V> implements Query<KeyValueIterator<K, ValueAndTimestamp<V>>> { ... /** * Interactive range query using a lower and upper bound to filter the keys returned. * @param lower The key that specifies the lower bound of the rangequery = TimestampRangeQuery.withRange(lower.orElse(null), upper.orElse(null)); final StateQueryRequest<KeyValueIterator<Integer, ValueAndTimestamp<V>>> request = inStore(STORE_NAME) .withQuery(query) * @param upper The key that specifies the upper bound of the range * @param .withPartitions(mkSet(0, 1))<K> The key type * @param <V> The value type */ public static <K, V> TimestampedRangeQuery<K, V> .withPositionBound(PositionBound.at(INPUT_POSITION)); withRange(final K lower, final K upper) /** * Determines finalif StateQueryResult<KeyValueIterator<Integer,the ValueAndTimestamp<V>>>serialized result = byte[] of the keys in ascending or descending or unordered order. * IntegrationTestUtils.iqv2WaitForResult(kafkaStreams, request); ... final Map<Integer, QueryResult<KeyValueIterator<Integer, ValueAndTimestamp<V>>>> queryResult = result.getPartitionResults()Order is based on the serialized byte[] of the keys, not the 'logical' key order. * @return return the order of return records base on the serialized byte[] of the keys (can be unordered, or in ascending, or in descending order). */ public ResultOrder resultOrder() { return order; } /** * Set the query to return the serialized byte[] of the keys in descending order... * Order is based on the serialized byte[] of the keys, not the 'logical' key order. * @return a trynew (finalRangeQuery KeyValueIterator<Integer,instance ValueAndTimestamp<V>>with iteratordescending =flag queryResult.get(partition).getResult()) { set. */ public TimestampedRangeQuery<K, V> withDescendingKeys() { return whilenew (iterator.hasNext()) {TimestampedRangeQuery<>(this.lower, this.upper, ResultOrder.DESCENDING); } /** * Set the query to return the serialized byte[] of the keys in Ascending actualValue.add(valueExtactor.apply(iterator.next().value));order. * Order is based on the serialized byte[] of the keys, not the 'logical' key }order. * @return a new RangeQuery instance with ascending flag } set. */ public TimestampedRangeQuery<K, V> withAscendingKeys() { return new TimestampedRangeQuery<>(this.lower, this.upper, ResultOrder.ASCENDING); } } |
According to KIP-968, we introduce a public enum ResultOrder.
ResultOrder enum
It helps with specifying the order of the returned results by the query.
Code Block | ||||
---|---|---|---|---|
| ||||
package org.apache.kafka.streams.query; public enum ResultOrder { }ANY, } |
Rejected Alternatives
Initially, our approach was to directly use TimestampKeyQuery and TimestampRangeQuery within each store. This implied that every store would return a ValueAndTimestamp<byte[]>. However, this method introduced complexities due to type transformations.
Given that data within stores under the metered store is typically formatted as <Byte, byte[]>, we would need to wrap the byte[] into ValueAndTimestamp<byte[]> to produce the desired output.
ASCENDING,
DESCENDING
} |
Compatibility, Deprecation, and Migration Plan
- Changing the semantics of existing
KeyQuery
andRangeQuery
is a breaking change. However, both classes are marked as@Evolving`
and thus a breaking change in a minor release is allowed without a deprecation period. Given that IQv2 is not yet widely adopted, we believe it’s cleaner to make this breaking change right away. - Adding new query types does not imply any compatibility concerns.
Test Plan
To ensure the robustness and accuracy of our new query types, TimestampedKeyQuery
and TimestampedRangeQuery
, it's essential to have thorough test coverage. With that in mind, we propose the creation of two specific test methods:
shouldHandleTimestampedKeyQuery
: This test method will validate the functionality of TimestampedKeyQuery
, ensuring it consistently returns ValueAndTimestamp<V>
as expected.
shouldHandleTimestampedRangeQuery
: This method is tailored to verify the TimestampedRangeQuery
, ensuring that it correctly returns a KeyValueIterator<K, ValueAndTimestamp<V>>
.
We will focus on conducting a detailed test for shouldHandleTimestampedRangeQuery
.
Rejected Alternatives
The alternative would be to deprecate the existing KeyQuery
and RangeQuery
and add new query types that always return plain value. However, it seems to introduce unnecessary “deprecation noise”, and it would be hard to find good names for these newly added query types. Making a semantics change on the existing queries allows us to keep the existing namesTo streamline this, we opted to leverage existing methods to fetch the byte[] directly, which already encapsulates both value and timestamp. We then perform the deserialization at the meteredTimestampKeyValueStore level, the outermost layer, ensuring that the final output is ValueAndTimestamp<V>.