1. A polylog competitive algorithm for the k-server problem Nikhil Bansal (Eindhoven) Niv Buchbinder (Open Univ., Israel) Aleksander Madry (MIT) Seffi Naor (Technion)

2. The k-server Problem 1 2 3 Move Closest Sever Algorithm

3. The Paging/Caching Problem Set of pages {1,2,…,n}, cache of size k<n. Request sequence of pages 1, 6, 4, 1, 4, 7, 6, 1, 3, … a) If requested page already in cache, no penalty. b) Else, cache miss. Need to fetch page in cache (possibly) evicting some other page. Goal: Minimize the number of cache misses. Paging: K-server on the uniform metric. (Server on location p = page p in cache) 1 n...

4. Previous Results: Paging Paging (Deterministic) [Sleator Tarjan 85]: Any deterministic algorithm >= k-competitive. LRU is k-competitive (also other algorithms) Paging (Randomized): Rand. Marking O(log k) [Fiat, Karp, Luby, McGeoch, Sleator, Young 91]. Lower bound H k [Fiat et al. 91], tight results known.

5. K-server conjecture [Manasse-McGeoch-Sleator ’88]: There exists k competitive algorithm on any metric space. Initially no f(k) guarantee. Fiat-Rababi-Ravid’90: exp(k log k) … Koutsoupias-Papadimitriou’95: 2k-1 Chrobak-Larmore’91: k for trees.

6. Randomized k-server Conjecture There is an O(log k) competitive algorithm for any metric. Uniform Metric: log k Polylog for very special cases (uniform-like) Line: n 2/3 [Csaba-Lodha’06] exp(O(log n) 1/2 ) [Bansal-Buchbinder-Naor’10] Depth 2-tree: No o(k) guarantee

7. Our Result Thm: There is an O(log 2 k log 3 n) competitive* algorithm for k-server on any metric with n points. * Hiding some log log n terms

8. Our Approach Hierarchically Separated Trees (HSTs) [Bartal 96]. Any Metric space Problems on HST often reduced to Uniform metrics. [Bartal-Blum-Burch-Tomkins 97, Kleinberg-Tardos 01, …] Allocation Problem (uniform metrics): [Cote-Meyerson-Poplawski’08] (We work with a weaker “fractional” allocation problem) O(log n)

9. Outline Introduction HSTs + Allocation Problem Fractional view of Randomized Algorithms Fractional Allocation Problem

10. Designing Algorithm on HST d+1 level HST

11. Allocation Problem

12. Allocation to k-server

13. Outline Introduction Allocation Problem Fractional view of Randomized Algorithms Fractional Allocation Algorithm

14. Fractional View of Randomized Algorithms To specify a randomized algorithm: i) Prob. distribution on states at time t. ii) How it changes at time t+1. Fractional view: Just specify some marginals. Eg. Paging, actual algorithm = distribution over k-tuples but, Fractional: p 1,…,p n s.t. p 1 + …+ p n = k Cost: If p 1,…,p n changes to q 1,…,q n, pay  i |p i – q i | Suffices: Fractional Paging -> Randomized Paging (2x loss)

15. Fractional Allocation Problem

16. A gap example Allocation Problem on 2 points Requests alternate on locations. Left: (1,1,…,1,0) Right: (1,0,…,0,0) Any integral solution must pay  (T) over T steps. Claim: Fractional Algorithm pays only T/(2k). Left: 0 servers w/p 1/k, and k servers w/p 1-1/k Right: has 1 server w/p 1. No move cost. Hit cost of 1/k on left requests. LeftRight

17. Main Steps

18. A word about Fractional Allocation

19. Concluding Remarks Removing dependence of depth on aspect ratio. Thm: HST -> Weighted HST with O(log n) depth. Extend Allocation to weighted star. Main question: Can we remove dependence on n. 1. Metric -> HST 2. But even on HST (lose depth of HST) 12 48

20. Thank you