Public cloud providers seek to meet stringent performance requirements and low hardware cost. A key driver of performance and cost is main memory. Memory pooling promises to improve DRAM utilization and thereby reduce costs. However, pooling is challenging under cloud performance requirements. This paper proposes Pond, the first memory pooling system that both meets cloud performance goals and significantly reduces DRAM cost. Pond builds on the Compute Express Link (CXL) standard for load/store access to pool memory and two key insights. First, our analysis of cloud production traces shows that pooling across 8-16 sockets is enough to achieve most of the benefits. This enables a small-pool design with low access latency. Second, it is possible to create machine learning models that can accurately predict how much local and pool memory to allocate to a virtual machine (VM) to resemble same-NUMA-node memory performance. Our evaluation with 158 workloads shows that Pond reduces DRAM costs by 7% with performance within 1-5% of same-NUMA-node VM allocations.

Cloud providers typically use air-based solutions for cooling servers in datacenters. However, increasing transistor counts and the end of Dennard scaling will result in chips with thermal design power that exceeds the capabilities of air cooling in the near future. Consequently, providers have started to explore liquid cooling solutions (e.g., cold plates, immersion cooling) for the most power-hungry workloads. By keeping the servers cooler, these new solutions enable providers to operate server components beyond the normal frequency range (i.e., overclocking them) all the time. Still, providers must tradeoff the increase in performance via overclocking with its higher power draw and any component reliability implications.In this paper, we argue that two-phase immersion cooling (2PIC) is the most promising technology, and build three prototype 2PIC tanks. Given the benefits of 2PIC, we characterize the impact of overclocking on performance, power, and reliability. Moreover, we propose several new scenarios for taking advantage of overclocking in cloud platforms, including oversubscribing servers and virtual machine (VM) auto-scaling. For the auto-scaling scenario, we build a system that leverages overclocking for either hiding the latency of VM creation or postponing the VM creations in the hopes of not needing them. Using realistic cloud workloads running on a tank prototype, we show that overclocking can improve performance by 20%, increase VM packing density by 20%, and improve tail latency in auto-scaling scenarios by 54%. The combination of 2PIC and overclocking can reduce platform cost by up to 13% compared to air cooling.

Cloud providers, like Amazon and Microsoft, must guarantee high availability for a large fraction of their workloads. For this reason, they build datacenters with redundant infrastructures for power delivery and cooling. Typically, the redundant resources are reserved for use only during infrastructure failure or maintenance events, so that workload performance and availability do not suffer. Unfortunately, the reserved resources also produce lower power utilization and, consequently, require more datacenters to be built. To address these problems, in this paper we propose "zero-reserved-power" datacenters and the Flex system to ensure that workloads still receive their desired performance and availability. Flex leverages the existence of software-redundant workloads that can tolerate lower infrastructure availability, while imposing minimal (if any) performance degradation for those that require high infrastructure availability. Flex mainly comprises (1) a new offline workload placement policy that reduces stranded power while ensuring safety during failure or maintenance events, and (2) a distributed system that monitors for failures and quickly reduces the power draw while respecting the workloads’ requirements, when it detects a failure. Our evaluation shows that Flex produces less than 5% stranded power and increases the number of deployed servers by up to 33%, which translates to hundreds of millions of dollars in construction cost savings per datacenter site. We end the paper with lessons from our experience bringing Flex to production in Microsoft’s datacenters.

Cloud research to date has lacked data on the characteristics of the production virtual machine (VM) workloads of large cloud providers. A thorough understanding of these characteristics can inform the providers' resource management systems, e.g. VM scheduler, power manager, server health manager. In this paper, we first introduce an extensive characterization of Microsoft Azure's VM workload, including distributions of the VMs' lifetime, deployment size, and resource consumption. We then show that certain VM behaviors are fairly consistent over multiple lifetimes, i.e. history is an accurate predictor of future behavior. Based on this observation, we next introduce Resource Central (RC), a system that collects VM telemetry, learns these behaviors offline, and provides predictions online to various resource managers via a general client-side library. As an example of RC's online use, we modify Azure's VM scheduler to leverage predictions in oversubscribing servers (with oversubscribable VM types), while retaining high VM performance. Using real VM traces, we then show that the prediction-informed schedules increase utilization and prevent physical resource exhaustion. We conclude that providers can exploit their workloads' characteristics and machine learning to improve resource management substantially.

Social content, such as Twitter updates, often have the quickest first-hand reports of news events, as well as numerous commentaries that are indicative of public view of such events. As such, social updates provide a good complement to professionally written news articles. In this paper we consider the problem of automatically annotating news stories with social updates (tweets), at a news website serving high volume of pageviews. The high rate of both the pageviews (millions to billions a day) and of the incoming tweets (more than 100 millions a day) make real-time indexing of tweets ineffective, as this requires an index that is both queried and updated extremely frequently. The rate of tweet updates makes caching techniques almost unusable since the cache would become stale very quickly. We propose a novel architecture where each story is treated as a subscription for tweets relevant to the story's content, and new algorithms that efficiently match tweets to stories, proactively maintaining the top-k tweets for each story. Such top-k pub-sub consumes only a small fraction of the resource cost of alternative solutions, and can be applicable to other large scale content-based publish-subscribe problems. We demonstrate the effectiveness of our approach on realworld data: a corpus of news stories from Yahoo! News and a log of Twitter updates.

Many large-scale Web applications that require ranked top-k retrieval are implemented using inverted indices. An inverted index represents a sparse term-document matrix, where non-zero elements indicate the strength of term-document associations. In this work, we present an approach for lossless compression of inverted indices. Our approach maps terms in a document corpus to a new term space in order to reduce the number of non-zero elements in the term-document matrix, resulting in a more compact inverted index. We formulate the problem of selecting a new term space as a matrix factorization problem, and prove that finding the optimal solution is an NP-hard problem. We develop a greedy algorithm for finding an approximate solution. A side effect of our approach is increasing the number of terms in the index, which may negatively affect query evaluation performance. To eliminate such effect, we develop a methodology for modifying query evaluation algorithms by exploiting specific properties of our compression approach.

Precomputation of common term co-occurrences has been successfully applied to improve query performance in large scale search engines based on inverted indexes. The results of such precomputations are traditionally stored as additional posting lists in the index. During query evaluation, these precomputed lists are used to reduce the number of query terms, as the results for multiple terms can be accessed through a single precomputed list. In this paper, we expand this paradigm by considering an alternative method for storing term co-occurrences in inverted indexes. For a selected set of terms in the index, we store bitmaps that encode term co-occurrences. A bitmap of size k for term t augments each posting to store the co-occurrences of t with k other terms, across every document in the index. At query evaluation, size k bitmaps can be used to answer queries that involve any of the 2^k combinations of the additional terms. In contrast, a precomputed list, although typically shorter, can only be used to evaluate queries containing all of its terms. We evaluate the bitmaps technique we propose, and the baseline of adding precomputed posting lists and show that they are complementary, as they capture different aspects of the query evaluation cost. We perform an experimental evaluation on the TREC WT10g corpus and show that a hybrid strategy combining both methods significantly lowers the cost of query evaluation compared to each method separately.

We introduce the problem of evaluating graph constraints in content-based publish/subscribe (pub/sub) systems. This problem formulation extends traditional content-based pub/sub systems in the following manner: publishers and subscribers are connected via a (logical) directed graph G with node and edge constraints, which limits the set of valid paths between them. Such graph constraints can be used to model a Web advertising exchange (where there may be restrictions on how advertising networks can connect advertisers and publishers) and content delivery problems in social networks (where there may be restrictions on how information can be shared via the social graph). In this context, we develop efficient algorithms for evaluating graph constraints over arbitrary directed graphs G. We also present experimental results that demonstrate the effectiveness and scalability of the proposed algorithms using a realistic dataset from Yahoo!'s Web advertising exchange.

Sponsored search is one of the major sources of revenue for search engines on the World Wide Web. It has been observed that while showing ads for every query maximizes short-term revenue, irrelevant ads lead to poor user experience and less revenue in the long-term. Hence, it is in search engines' interest to place ads only for queries that are likely to attract ad-clicks. Many algorithms for estimating query advertisability exist in literature, but most of these methods have been proposed for and tested on the frequent or "head" queries. Since query frequencies on search engine are known to be distributed as a power-law, this leaves a huge fraction of the queries uncovered.

The problem of efficiently evaluating a large collection of complex Boolean expressions - beyond simple conjunctions and Disjunctive/Conjunctive Normal Forms (DNF/CNF) - occurs in many emerging online advertising applications such as advertising exchanges and automatic targeting. The simple solution of normalizing complex Boolean expressions to DNF or CNF form, and then using existing methods for evaluating such expressions is not always effective because of the exponential blow-up in the size of expressions due to normalization. We thus propose a novel method for evaluating complex expressions, which leverages existing techniques for evaluating leaf-level conjunctions, and then uses a bottom-up evaluation technique to only process the relevant parts of the complex expressions that contain the matching conjunctions. We develop two such bottom-up evaluation techniques, one based on Dewey IDs and another based on mapping Boolean expressions to one-dimensional intervals. Our experimental evaluation based on data obtained from an online advertising exchange shows that the proposed techniques are efficient and scalable, both with respect to space usage as well as evaluation time.

We explore the use of the landing page content in sponsored search ad selection. Specifically, we compare the use of the ad's intrinsic content to augmenting the ad with the whole, or parts, of the landing page. We explore two types of extractive summarization techniques to select useful regions from the landing pages: out-of-context and in-context methods. Out-of-context methods select salient regions from the landing page by analyzing the content alone, without taking into account the ad associated with the landing page. In-context methods use the ad context (including its title, creative, and bid phrases) to help identify regions of the landing page that should be used by the ad selection engine. In addition, we introduce a simple yet effective unsupervised algorithm to enrich the ad context to further improve the ad selection. Experimental evaluation confirms that the use of landing pages can significantly improve the quality of ad selection. We also find that our extractive summarization techniques reduce the size of landing pages substantially, while retaining or even improving the performance of ad retrieval over the method that utilize the entire landing page.

Contextual advertising (also called content match) refers to the placement of small textual ads within the content of a generic web page. It has become a significant source of revenue for publishers ranging from individual bloggers to major newspapers. At the same time it is an important way for advertisers to reach their intended audience. This reach depends on the total number of exposures of the ad (impressions) and its click-through-rate (CTR) that can be viewed as the probability of an end-user clicking on the ad when shown. These two orthogonal, critical factors are both difficult to estimate and even individually can still be very informative and useful in planning and budgeting advertising campaigns.

We propose a methodology for building a robust query classification system that can identify thousands of query classes, while dealing in real time with the query volume of a commercial Web search engine. We use a pseudo relevance feedback technique: given a query, we determine its topic by classifying the Web search results retrieved by the query. Motivated by the needs of search advertising, we primarily focus on rare queries, which are the hardest from the point of view of machine learning, yet in aggregate account for a considerable fraction of search engine traffic. Empirical evaluation confirms that our methodology yields a considerably higher classification accuracy than previously reported. We believe that the proposed methodology will lead to better matching of online ads to rare queries and overall to a better user experience.

Web textual advertising can be interpreted as a search problem over the corpus of ads available for display in a particular context. In contrast to conventional information retrieval systems, which always return results if the corpus contains any documents lexically related to the query, in Web advertising it is acceptable, and occasionally even desirable, not to show any results. When no ads are relevant to the user's interests, then showing irrelevant ads should be avoided since they annoy the user and produce no economic benefit. In this paper we pose a decision problem to swing, that is, whether or not to show any of the ads for the incoming request. We propose two methods for addressing this problem, a simple thresholding approach and a machine learning approach, which collectively analyzes the set of candidate ads augmented with external knowledge. Our experimental evaluation, based on over 28,000 editorial judgments, shows that we are able to predict, with high accuracy, when to swing for both content match and sponsored search advertising.

The business of Web search, a $10 billion industry, relies heavily on sponsored search, whereas a few carefully-selected paid advertisements are displayed alongside algorithmic search results. A key technical challenge in sponsored search is to select ads that are relevant for the user's query. Identifying relevant ads is challenging because queries are usually very short, and because users, consciously or not, choose terms intended to lead to optimal Web search results and not to optimal ads. Furthermore, the ads themselves are short and usually formulated to capture the reader's attention rather than to facilitate query matching.

In contextual advertising, estimating the number of impres-sions of an ad is critical in planning and budgeting adver-tising campaigns. However, producing this forecast, even within large margins of error, is quite challenging. We attack this problem by simulating the presence of a given ad with its associated bid over historical data, involving billions of impressions. This apparently enormous computational task is reduced to a search task involving only the set of distinct pages in the data. Furthermore the search is made more effi-cient using a two-level search process. Experimental results show that our approach can accurately forecast the expected number of impressions of contextual ads in real time.

In this paper we propose a novel document retrieval model in which text queries are augmented with multi-dimensional taxonomy restrictions. These restrictions may be relaxed at a cost to result quality. This new model may be applicable in many arenas, including multifaceted, product, and local search, where documents are augmented with hierarchical metadata such as topic or location. We present efficient algorithms for indexing and query processing in this new retrieval model. We decompose query processing into two sub-problems: first, an online search problem to determine the correct overall level of relaxation cost that must be incurred to generate the top k results; and second, a budgeted relaxation search problem in which all results at a particular relaxation cost must be produced at minimal cost. We show the latter problem is solvable exactly in two hierarchical dimensions, is NP-hard in three or more dimensions, but admits efficient approximation algorithms with provable guarantees. We present experimental results evaluating our algorithms on both synthetic and real data, showing order of magnitude improvements over the baseline algorithm.

We propose a methodology for building a practical robust query classification system that can identify thousands of query classes with reasonable accuracy, while dealing in real-time with the query volume of a commercial web search engine. We use a blind feedback technique: given a query, we determine its topic by classifying the web search results retrieved by the query. Motivated by the needs of search advertising, we primarily focus on rare queries, which are the hardest from the point of view of machine learning, yet in aggregation account for a considerable fraction of search engine traffic. Empirical evaluation confirms that our methodology yields a considerably higher classification accuracy than previously reported. We believe that the proposed methodology will lead to better matching of online ads to rare queries and overall to a better user experience.

Contextual advertising or Context Match (CM) refers to the placement of commercial textual advertisements within the content of a generic web page, while Sponsored Search (SS) advertising consists in placing ads on result pages from a web search engine, with ads driven by the originating query. In CM there is usually an intermediary commercial ad-network entity in charge of optimizing the ad selection with the twin goal of increasing revenue (shared between the publisher and the ad-network) and improving the user experience. With these goals in mind it is preferable to have ads relevant to the page content, rather than generic ads. The SS market developed quicker than the CM market, and most textual ads are still characterized by "bid phrases" representing those queries where the advertisers would like to have their ad displayed. Hence, the first technologies for CM have relied on previous solutions for SS, by simply extracting one or more phrases from the given page content, and displaying ads corresponding to searches on these phrases, in a purely syntactic approach. However, due to the vagaries of phrase extraction, and the lack of context, this approach leads to many irrelevant ads. To overcome this problem, we propose a system for contextual ad matching based on a combination of semantic and syntactic features.

We consider the problem of estimating the size of a collection of documents using only a standard query interface. Our main idea is to construct an unbiased and low-variance estimator that can closely approximate the size of any set of documents defined by certain conditions, including that each document in the set must match at least one query from a uniformly sampleable query pool of known size, fixed in advance.Using this basic estimator, we propose two approaches to estimating corpus size. The first approach requires a uniform random sample of documents from the corpus. The second approach avoids this notoriously difficult sample generation problem, and instead uses two fairly uncorrelated sets of terms as query pools; the accuracy of the second approach depends on the degree of correlation among the two sets of terms.Experiments on a large TREC collection and on three major search engines demonstrates the effectiveness of our algorithms.

A major difference between corporate intranets and the Internet is that in intranets the barrier for users to create web pages is much higher. This limits the amount and quality of anchor text, one of the major factors used by Internet search engines, making intranet search more difficult. The social phenomenon at play also means that spam is relatively rare. Both on the Internet and in intranets, users are often willing to cooperate with the search engine in improving the search experience. These characteristics naturally lead to considering using user feedback to improve search quality in intranets. In this paper we show how a particular form of feedback, namely user annotations, can be used to improve the quality of intranet search. An annotation is a short description of the contents of a web page, which can be considered a substitute for anchor text. We propose two ways to obtain user annotations, using explicit and implicit feedback, and show how they can be integrated into a search engine. Preliminary experiments on the IBM intranet demonstrate that using annotations improves the search quality.

Maintaining strict static score order of inverted lists is a heuristic used by search engines to improve the quality of query results when the entire inverted lists cannot be processed. This heuristic, however, increases the cost of index generation and requires complex index build algorithms. In this paper, we study a new index organization based on static score bucketing. We show that this new technique significantly improves in index build performance while having minimal impact on the quality of search results.

Holistic twig join algorithms represent the state of the art for evaluating path expressions in XML queries. Using inverted indexes on XML elements, holistic twig joins move a set of index cursors in a coordinated way to quickly find structural matches. Because each cursor move can trigger I/O, the performance of a holistic twig join is largely determined by how many cursor moves it makes, yet, surprisingly, existing join algorithms have not been optimized along these lines. In this paper, we describe TwigOptimal, a new holistic twig join algorithm with optimal cursor movement. We sketch the proof of TwigOptimal's optimality, and describe how TwigOptimal can use information in the return clause of XQuery to boost its performance. Finally, experimental results are presented, showing TwigOptimal's superiority over existing holistic twig join algorithms.

All known algorithms for evaluating advanced XPath queries (e.g., ones with predicates or with closure axes) on XML streams employ buffers to temporarily store fragments of the document stream. In many cases, these buffers grow very large and constitute a major memory bottleneck. In this paper, we identify two broad classes of evaluation problems that independently necessitate the use of large memory buffers in evaluation of queries over XML streams: (1) full-fledged evaluation (as opposed to just filtering) of queries with predicates; (2) evaluation (whether full-fledged or filtering) of queries with "multi-variate" predicates.We prove quantitative lower bounds on the amount of memory required in each of these scenarios. The bounds are stated in terms of novel document properties that we define. We show that these scenarios, in combination with query evaluation over recursive documents, cover the cases in which large buffers are required. Finally, we present algorithms that match the lower bounds for an important fragment of XPath.

Structural joins are a fundamental operation in XML query processing and a large body of work has focused on index-based algorithms for executing them. In this paper, we describe how two well-known index features -- path indices and ancestor information -- can be combined in a novel way to replace one or more of the physical index cursors in a structural join with virtual cursors. The position of a virtual cursor is derived from the path and ancestor information of a physical cursor. Implementation results are provided to show that, by eliminating index I/O, virtual cursors can improve the performance of structural joins by an order of magnitude or more.

The important challenge of evaluating XPath queries over XML streams has sparked much interest in the past two years, A number of algorithms have been proposed, supporting wider fragments of the query language, and exhibiting better performance and memory utilization. Nevertheless, all the algorithms known to date use a prohibitively large amount of memory for certain types of queries. A natural question then is whether this memory bottleneck is inherent or just an artifact of the proposed algorithms.In this paper we initiate the first systematic and theoretical study of lower bounds on the amount of memory required to evaluate XPath queries over XML streams. We present a general lower bound technique, which given a query, specifies the minimum amount of memory that any algorithm evaluating the query on a stream would need to incur. The lower bounds are stated in terms of new graph-theoretic properties of queries. The proof is based on tools from communication complexity.We then exploit insights learned from the lower bounds to obtain a new algorithm for XPath evaluation on streams. The algorithm uses space close to the optimum. Our algorithm deviates from the standard paradigm of using automata or transducers, thereby avoiding the need to store large transition tables.

We present a streaming algorithm for evaluating XPath expressions that use backward axes (parent and ancestor) and forward axes in a single document-order traversal of an XML document. Other streaming XPath processors handle only forward axes. We show through experiments that our algorithm significantly outperforms (by more than a factor of two) a traditional nonstreaming XPath engine. Furthermore, our algorithm scales better because it retains only the relevant portions of the input document in memory. Our engine successfully processes documents over 1GB in size, whereas the traditional XPath engine degrades considerably in performance for documents over 100 MB in size and fails to complete for documents of size over 200 MB.

This paper proposes a flexible architecture for the creation of Internet auctions. It allows the custom definition of the auction parameters, and provides a decentralized control of the auction process. Auction policies are defined as laws in the Law Governed Interaction (LGI) paradigm. Each of these laws specifies not only the auction algorithm itself (e.g. open-cry, dutch, etc.) but also how to handle the other parameters usually involved in the online auctions, such as certification, auditioning, and treatment of complaints. LGI is used to enforce the rules established in the auction policy within the agents involved in the process. After the agents find out about the actions, they interact in a peer-to-peer communication protocol, reducing the role of the centralized auction room to an advertising registry, and taking profit of the distributed nature of the Internet to conduct the auction. The paper presents an example of an auction law, illustrating the use of the proposed architecture.