*   <li>Access to all collection-based utilities via {@link #asList} (though at the cost of
 *       allocating garbage).
 * </ul>
 *
 * <p>Disadvantages compared to {@code double[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).
 *   <li><i>Some</i> construction use cases force the data to be copied (though several construction
 *       APIs are offered that don't).

the value in the form of a `Handle[T]`.
Two `Handle[T]` are equal if and only if the values used to produce the
handles are equal, allowing programs to deduplicate values and reduce
their memory footprint.
Comparing two `Handle[T]` values is efficient, reducing down to a simple

 *   <li>Access to all collection-based utilities via {@link #asList} (though at the cost of
 *       allocating garbage).
 * </ul>
 *
 * <p>Disadvantages compared to {@code long[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).
 *   <li><i>Some</i> construction use cases force the data to be copied (though several construction
 *       APIs are offered that don't).

 * lock, but also maximizes memory footprint. On the other extreme, one could use a single lock for
 * all tasks, which minimizes memory footprint but also minimizes concurrency. Instead of choosing
 * either of these extremes, {@code Striped} allows the user to trade between required concurrency
 * and memory footprint. For example, if a set of tasks are CPU-bound, one could easily create a

 * traditional {@code LinkedHashMap} implementation which stores keys and count values as map
 * entries, {@code ObjectCountLinkedHashMap} minimizes object allocation and reduces memory
 * footprint.
 */
@GwtCompatible(serializable = true, emulated = true)
@ElementTypesAreNonnullByDefault
class ObjectCountLinkedHashMap<K extends @Nullable Object> extends ObjectCountHashMap<K> {

Ztunnel is not general purpose; it has an extremely tight goal.
We can exploit this to make a more efficient protocol, which is critical to achieve our resource footprint goals.

For example, configuring Istio mTLS in Envoy takes roughly 50 lines of JSON (it is in Protobuf, of course, but still relevant).

 * a traditional {@code HashMap} implementation which stores keys and count values as map entries,
 * {@code ObjectCountHashMap} minimizes object allocation and reduces memory footprint.
 *
 * <p>In the absence of element deletions, this will iterate over elements in insertion order.
 */
@GwtCompatible(serializable = true, emulated = true)
@ElementTypesAreNonnullByDefault

   * hashmaps. Using a static ThreadLocal to avoid that overhead would mean that different
   * ExecutionSequencer objects interfere with each other, which would be undesirable, in addition
   * to increasing the memory footprint of every thread that interacted with it. In order to release
   * entries in thread-specific maps when the ThreadLocal object itself is no longer referenced,

   * E} is an {@link Enum} type, use {@link EnumSet#noneOf} instead. Otherwise, strongly consider
   * using a {@code LinkedHashSet} instead, at the cost of increased memory footprint, to get
   * deterministic iteration behavior.
   *
   * <p><b>Note:</b> this method is now unnecessary and should be treated as deprecated. Instead,
   * use the {@code HashSet} constructor directly, taking advantage of <a

components normalized by 1000 requests/second. When idle or low traffic, this chart will be blank. The curve for istio-proxy refers to the services sidecars only.\n- **vCPU:** vCPU utilization by Istio components, not normalized.\n- **Memory:** memory footprint for the components. Telemetry and policy are normalized by 1k rps, and no data is shown  when there is no traffic. For ingress and istio-proxy, the data is per instance.\n- **Bytes transferred / sec:** shows the number of bytes flowing through each...