Алексей Петров "PHP at Scale: Knowing enough to be dangerous!"
0 likes1,072 views
PHP at Scale: Knowing enough to be dangerous! by Oleksii Petrov discusses how to scale PHP applications. It covers strategies like caching, queueing, read/write splitting, and sharding. It also discusses using load balancers and choosing the right database. The key is to improve system metrics without dramatically changing the system. Scaling is predefined by your stack and architecture. Performance comes from optimizations everywhere, not just PHP. Being distributed is very challenging.
Алексей Петров "PHP at Scale: Knowing enough to be dangerous!"
- 1. PHP at Scale: Knowing enough to be dangerous! Oleksii Petrov Skelia Ukraine / ETWater
- 2. Who am I System Architect Team Lead PHP Developer Find me on @alexhelkar alexhelkar https://github.com/alexhelkar
- 3. What does it mean "to scale"?
- 5. Or does it mean?
- 7. "To scale" Improve system metrics without changing the system* * (dramaticaly)
- 8. Scale roadmap is predefined by your stack and software architecture
- 9. Let's go!
- 11. PHP 7 facts isset 1.55 times faster than array_key_exists is_file 26 times faster than file_exists single quotes slower that double quotes instanceof faster is_a etc. Code performance?
- 12. Blackfire.io & Symfony Blog Current results: Total request time: 18.4 ms file_exists called: 7 times Exec. time for file_exists: 460 µs Expected results: Exec/ time: ~17.6 µs Total request time: ~18.3996 ms
- 14. Why a system is slow?
- 17. Caching
- 19. DB Cache
- 21. Web Cache
- 22. Cache: Rule of thumb IF Application is SLOW Enable Caching IF Application has GLITCHES Disable Caching
- 23. Queueing
- 24. Queue as The Equalizer
- 25. Queue as The Equalizer
- 26. Problems? Delays
- 27. Queue as Load Balancer
- 32. Galera Replication aka sync replication
- 34. What is IoT?
- 35. IoT is like
- 36. Canvas REST API Weather Data Avg. Request Size 325b
- 39. Tasks Create platform setup Gather performance metrics Calculate budgets
- 40. Plan 1: 86k req/day (1 rps)
- 41. Plan 2: 8.64M req/day (100 rps)
- 42. Setup: 20$
- 43. Response Time
- 44. Limitation
- 47. Probability? Head: 50% | Tails: 50%
- 48. Probability of 2 consecutive Heads? 25%
- 49. Probability of 3 consecutive Heads? 12.5% 0.5^3
- 50. Servers reliability 1 Server 99% 10 Servers 90.4% 100 Servers 36.6%
- 51. Distributed System High Availability Any should consider
- 52. Plan 4: 43.2M req/day (500rps)
- 53. Setup Variant 1: 380$
- 54. Quantiles
- 55. Setup Variant 2: 150$
- 56. Quantiles
- 57. Mircoservices 230 Reasons to choose 380$ - 150$ = 230$
- 58. Plan 5: 84.6M req/day (1k rps)
- 59. Plan 5: 84.6M req/day (1k rps)
- 60. Plan 6: 432M req/day (5k rps)
- 61. Started to think about bulk inserts?
- 62. Plan 7: 1.26B req/day (15k rps)
- 63. Quantiles
- 64. TCP/IP
- 65. TCP Flow
- 66. Packets Rate
- 67. Packets Rate 1 Request 15k Requests ~10 Packets ~15k Packets
- 72. Why?
- 73. Load Balancer Trick 75k pack. 75k pack. LB Recieved: 150k packets
- 74. Plan 7: 50000 req/sec
- 76. Commodity Hardware $5 Server 1 CPU 20 CPU $100 Server $640 Server
- 80. DNS Cache
- 81. Rule of Thumb: Add LB
- 82. What if?
- 83. Floating IP
- 86. Mysql: Fabric (pre-alpha) <?php $mysqli = new mysqli("myapp", "user", "password", "database"); mysqlnd_ms_fabric_select_shard($mysqli, "test.fabrictest", 10); $mysqli->query("INSERT INTO fabrictest(id) VALUES (10)"); mysqlnd_ms_fabric_select_shard($mysqli, "test.fabrictest", 10); $mysqli->query("SELECT id FROM test WHERE id = 10"); http://php.net/manual/ru/mysqlnd- ms.quickstart.mysql_fabric.php
- 87. MariaDB: Spider
- 89. MongoDB Sharding
- 90. Apache Cassandra
- 91. RabbitMQ Hits 1 Million Messages Per Second. 32 machines https://blog.pivotal.io/pivotal/products/rab bitmq-hits-one-million-messages-per- second-on-google-compute-engine
- 92. Apache Kafka 2 Million Writes Per Second 3 machines https://engineering.linkedin.com/kafka/be nchmarking-apache-kafka-2-million-writes- second-three-cheap-machines
- 93. Instead Summary Scalability as afterwords doesn't work Database is not a queue Database is not a lock system Redis is not a queue Load balancers as a must Linear disk writes wins Perfomance in PHP is done by everything except PHP Being distributed is freaking hard