diff options
author | Alex Auvolat <alex@adnab.me> | 2021-02-21 14:30:26 +0100 |
---|---|---|
committer | Alex Auvolat <alex@adnab.me> | 2021-02-21 14:30:26 +0100 |
commit | 217b0dfd683cda0a082fb33636b38f0172a4af31 (patch) | |
tree | 81064bc739febb48e882eaeb0e578e21d410b635 /script | |
parent | 80892df8cce74211960c4259f48cd8c1bd5290be (diff) | |
download | garage-217b0dfd683cda0a082fb33636b38f0172a4af31.tar.gz garage-217b0dfd683cda0a082fb33636b38f0172a4af31.zip |
Add script to simulate different kinds of rings
Diffstat (limited to 'script')
-rw-r--r-- | script/simulate_ring.py | 120 |
1 files changed, 120 insertions, 0 deletions
diff --git a/script/simulate_ring.py b/script/simulate_ring.py new file mode 100644 index 00000000..fa875d4e --- /dev/null +++ b/script/simulate_ring.py @@ -0,0 +1,120 @@ +import hashlib +import bisect +import xxhash + + +def walk_ring_from_pos(tokens, dcs, start, rep): + ret = [] + ret_dcs = set() + delta = 0 + while len(ret) < rep: + i = (start + delta) % len(tokens) + delta = delta + 1 + + (token_k, token_dc, token_node) = tokens[i] + if token_dc not in ret_dcs: + ret_dcs |= set([token_dc]) + ret.append(token_node) + elif len(ret_dcs) == len(dcs) and token_node not in ret: + ret.append(token_node) + return ret + +def count_tokens_per_node(tokens): + tokens_of_node = {} + for _, _, token_node in tokens: + if token_node not in tokens_of_node: + tokens_of_node[token_node] = 0 + tokens_of_node[token_node] += 1 + print("#tokens per node:") + for node, ntok in sorted(list(tokens_of_node.items())): + print(node, ": ", ntok) + +def method1(nodes): + tokens = [] + dcs = set() + for (node_id, dc, n_tokens) in nodes: + dcs |= set([dc]) + for i in range(n_tokens): + token = hashlib.sha256(f"{node_id} {i}".encode('ascii')).hexdigest() + tokens.append((token, dc, node_id)) + tokens.sort(key=lambda tok: tok[0]) + + count_tokens_per_node(tokens) + + space_of_node = {} + + def walk_ring(v, rep): + i = bisect.bisect_left([tok for tok, _, _ in tokens], hashlib.sha256(v).hexdigest()) + return walk_ring_from_pos(tokens, dcs, i, rep) + + return walk_ring + + +def method2(nodes): + partition_bits = 10 + partitions = list(range(2**partition_bits)) + def partition_node(i): + h, hn, hndc = None, None, None + for (node_id, node_dc, n_tokens) in nodes: + for tok in range(n_tokens): + #hnode = hashlib.sha256(f"{i} {node_id} {tok}".encode('ascii')).hexdigest() + xxh = xxhash.xxh64() + xxh.update(f"partition {i} node {node_id} token {tok}".encode('ascii')) + hnode = xxh.digest() + if h is None or hnode < h: + h = hnode + hn = node_id + hndc = node_dc + return (i, hndc, hn) + + partition_nodes = [partition_node(i) for i in partitions] + count_tokens_per_node(partition_nodes) + + dcs = list(set(node_dc for _, node_dc, _ in nodes)) + + + def walk_ring(v, rep): + vh = hashlib.sha256(v).digest() + i = (vh[0]<<8 | vh[1]) % (2**partition_bits) + return walk_ring_from_pos(partition_nodes, dcs, i, rep) + + return walk_ring + + + +def evaluate_method(walk_ring): + node_data_counts = {} + for i in range(100000): + nodes = walk_ring(f"{i}".encode('ascii'), 3) + for n in nodes: + if n not in node_data_counts: + node_data_counts[n] = 0 + node_data_counts[n] += 1 + print("Number of data items per node:") + for n, v in sorted(list(node_data_counts.items())): + print(n, ": ", v) + + +if __name__ == "__main__": + + nodes = [('digitale', 'atuin', 10), + ('drosera', 'atuin', 10), + ('datura', 'atuin', 10), + ('io', 'jupiter', 20), + ('meta', 'pipo', 10), + ('mega', 'pipo', 10), + ('mina', 'pipo', 10), + #('moni', 'pipo', 10), + #('mimi', 'pipo', 10), + #('mesi', 'pipo', 10), + ] + + print("------") + print("method 1") + method1_walk_ring = method1(nodes) + evaluate_method(method1_walk_ring) + + print("------") + print("method 2") + method2_walk_ring = method2(nodes) + evaluate_method(method2_walk_ring) |